Royer ZVS Induction Heater

Published 18. January 2013. Updated 3. August 2021.


The Mazilli ZVS flyback driver is well-known throughout the high voltage community for its simplicity and ability to deliver 20-50 kV at high currents for a flyback transformer. About one and a half year ago, Marko from gave the circuit a comeback with it converted to a simple induction heater.

To explore all my induction heaters, including the Chinese 1800 Watt induction heater, check out my youtube playlist for all induction heater related projects: There is also a great thread on High Voltage Forum with details on many ZVS induction heaters and how to use them:


WARNING!: Working with electricity is dangerous, all information found on my site is for educational purpose and I accept no responsibility for others actions using the information found on this site.

Read this document about safety!



The MOSFETs used need a voltage rating about 4 times higher than the supply voltage and a on-resistance below 150 mΩ. In ZVS operation the switches see a voltage that is π times input voltage, so 4 times rating of input voltage leaves some head room for playing it safe.

If supply voltage gets over 40 VDC, consider using resistors between 470R-800R for the gates. Supply voltage needs to be minimum 12 VDC, lower than 470R gate resistors can be used in that case, if supply voltage dips under 10 VDC, there is a risk of MOSFETs failing from overheating by working only in the linear region or short circuit if one of them stops switching.

Supply voltage should not exceed 60 VDC, as this is very close to 200 VDC across the MOSFET. The internal construction of MOSFETs with a higher voltage rating makes them unsuitable for use in a self oscillating circuit like this Royer oscillator.

A MMC is made from 27 capacitors to avoid excessive heating in a single capacitor. The capacitors will still heat as massive current flows between the tank and work coil. To get a good result, a large tank capacitance is needed, if a capacitance lower than 4 uF is used, results might be disappointing. It is strongly advised to use a capacitor with made from polypropylene (MKP) or similar that can handle large RMS currents, it might even be necessary to water cool the capacitor too. A MMC as the one I use here can only withstand short run times and will even then heat up.

The value of the inductors are advised to be between 45 to 200 uH and depending on core material the number of turns varies a lot, use a LCR meter to check the values.

Water cooling of the work coil is a must! Even at just small runs with moderate power input as the ones I have conducted, the work coil would take damage from heat.


Voltage supply35 VDC smoothed with 40000 uF
MMC3 uF from 9 in parallel strings of 3x 2 275 VAC MKP X2 capacitors in series.
Power consumption650 Watt.
Best resultBetween red hot and white hot M10x20mm bolt





17th January 2013

I succeeded in putting the entire setup together from parts I have salvaged from old equipment, only the MOSFETs was bought new and used before.

The transformer takes 230 VAC in for 32 VAC out, properly around 700 VA transformer estimated from the core size. It is rectified with a 25 A bridge rectifier smoothed with 40000 uF capacitance from four electrolytic capacitors 70 VDC / 10000 uF each.

The inductors are made from ferrite transformer cores from old power supplies. 14 turns of 1,5 mm^2 gave approximately 130 uH inductance.

Two IRFP250N MOSFETs mounted on each their fairly small heat sink, but big enough for the circuit to run for a couple of minutes and only get a little above hand warm. The heat sinks are glued together with a piece of acrylic plastic in-between to insure electrical isolation between the two heat sinks.

The work coil is made from 5 turns of 8 mm copper tubing, giving approximately 0,477 uH. The MMC consists of 9 parallel strings of 3 in series Rifa 1 uF / 275 VAC MKP X2 capacitors for 3 uF. This gives a resonant frequency calculated to about 133 kHz.

Measurements during a run of heating a M10x20 mm bolt at 33 VDC in, 260 VAC at 2.5 A input into transformer.

Resonant frequency is measured to 106 kHz. The measured frequency is different from the calculated as the work piece will influence on the coils electromagnetic properties.

In the following oscilloscope screenshot:

Yellow: Inverter current, here measured to 10 Ampere.

Blue: Inverter voltage, here measured to 100 Volt.

In the following oscilloscope screenshot:

Yellow: Tank current, here measured to 200 Ampere.

Blue: Tank voltage, here measured to 100 Volt.

Three pieces of metal heated to what is possible with input voltage of 35 VDC.

First picture shows a M4x10 mm screw heated red hot at 260 Watt input power, second and third picture shows a M10x20 mm bolt heated red hot at 650 Watt input power.



A good and reliable oscillator as long as supply voltage is kept within safe area of operation for the MOSFETs and only short run times are used unless there is used good components and water cooling on work coil, MOSFETs and capacitors.

Further improvements in use as a heater / melter would be a higher supply voltage.



410 thoughts on “Royer ZVS Induction Heater”

  1. Pingback: Royer induction heater – first test | Kaizer Power Electronics

  2. hi, good project but i have a question : would it be ok if i would use 1n5352b instead on 1n5349?

  3. Hey Augustin, you would be driving the gates a little bit harder with 15V instead of 12V, it should work without much trouble, it is first when you get above 20V you could get in trouble.

  4. Hey i want do it for the exam of the 5th class… Is it good take the inductance from a pc power supply?

  5. Hello Marco

    The ferrite cores you find in a pc power supply is powdered iron cores and while they can be used, they will saturate faster and heat up more than a ferrite core would, if I remember correctly you need to have more turns on a iron powder core to get the same inductance, get a cheap LCR meter from hong kong over ebay if you do not have one.

    Kind regards

  6. Hi, im doing a similar project but my source power only gives 30Vdc & 6amps, and the voltajes drops to 14v….how much current gives your transformer??

    Another cuestion, how do you calculate the resonant frecuency???

  7. Hi Mads,

    Does the same circuit work for a pancake coil with 50uH?
    Have you created a simulation file e.g. in pspice with which I could try?

    Thank you for your help!


  8. Hello Verena

    Yes this circuit would also work with a pancake coil so you could basically use it as an induction stove, you would have to experiment with coupling between coil and work piece or pan!

    I do not have any simulations.

    Kind regards

  9. I have tested to follow your circuits.It could not work.
    I could noticed.I could see to short circuit of mosfet and zener diode,
    So,I would like to know resonal of fail circuit.and suggeust to me please.
    This is the detial of electronic devices:
    1) Transformer 24V (25A)
    2) Mosfet IRFP250N
    3) Resistor 470E/1W and 10K/1W
    4) Diode Bridge 25A
    5) Zener doide 1N5349B
    6) Capacitor 1u/400V (MKP)
    7) Inductor are made from teriod
    and the last I would like to know the diamiter size of working coil.
    Thank you very much.

  10. Hello prachum

    Your component list seems fine, should work with those items.

    Make sure you have the MOSFETs mounted correctly, have them on a heat sink, have them isolated with pads if they are mounted on the same heat sink. In the schematic, if lines are crossing, only the dots show a connection.

    The diameter of the work coil is 50 mm on the inside as I wound it around a 50 mm diameter pipe.

    Kind regards

  11. Hello Mads,
    I am no good at physics but I have a more general question aiming to keep my boat warm at low running cost.
    If the only power supply was only 12V through vehicle standard batteries and the job is to just heat a small place using a fan blower, then would that be technically possible? (altering specs and design of course)
    If the answer is yes, then how would the battery Amp hours needed compare with an equally powerful resistor that would provide the same heat??

  12. Hello George

    If you just need heating, I would look at a regular heater with a fan, there is now advantage in this circuit when you just want heat, you will only get as much energy in heat as energy you put into your circuit.

    Kind regards

  13. Thank you answering my stupid q.
    From it, am I assuming correctly that if we were to build a device with only one single ‘suitable’ shape resistor able to finally produce the same heat as yours then it would draw the same amps as your device??

  14. Hey George

    A induction heater works much different than just power dissipated in a resistive wire as you will find in a heater fan. For heating purposes, forget all about this circuit, if you want to heat up metal or melt metal, then this is the circuit for you.

    Kind regards

  15. I have a few questions for this circuit before I go an build it.

    1) How critical are the values of the Inductors?
    2) How critical are the values of the RURP880 diodes?
    3) I have a 12V 48A supply, will this be an issue for getting good results? Is there anyway to modify the circuit to take advantage of the 48A that I have available to me, since 12V isn’t that much?

    Thank you!

  16. Hey Austin

    I measured the values of the inductors with a cheap LCR meter, as long as they are in the range of some 10s uH, you should be alright.

    The important specifications of the diodes is voltage rating and speed, ultra-fast diodes should be used.

    The zener diodes over the gates of the MOSFETs are 12V, so with only a 12V supply you risk that it could stop oscillating if the voltage on the gate gets below 10V, either one MOSFET will stop switching or they will overheat from only working in the liniar region if they switch too slow.

    You can use a 12V supply, if you are absolutely sure it will not drop below, it sounds like a SMPS, so you properly need a DC-DC converter to get a higher voltage.

    Kind regards

  17. Okay, so I already have a pair of 10A 100uH Inductors, should those be fine?

    I also have some Ultrafast Diodes, but they are rated only for 600V, not 800V, will that be an issue?

    For the Zener diodes, since I will be using a 12V supply, could I use a slightly lower value Zener to prevent the stop of oscillation?

    Unfortunately a DC-DC converter is a bit out of my price range because I’m currently going to college and don’t have too much money to spend as it is.


  18. You can try with those inductors, I can not tell you if they will work or not.

    Remember that the MOSFETs only see 3,14 times the input voltage, so even with switching spikes, it might only be around 10 times input voltage that is the needed rating for the diodes.

    Use 12V zeners, used correctly they do not have the voltage drop of a normal diode, just be sure to watch your input voltage when running the circuit.

    Go to a nearby scrap yard and buy some old transformers, thats a cheap way to get transformers in the 1kVA range.

    Kind regards

  19. Hello

    Nice site, great projects!

    I would like to construct the powered-up version of this circuit
    (230VAC input, 1Phase, 50Hz, 16A max)

    What do you think of these components:
    – EMI filter, 13A fuses
    – Of-the-shelf triac power regulation (
    (only for the range 0-30% and a rapid turn towards 100%)

    – 25A full bridge rectifier, 450V capacitor bank (6000uF or more)
    – SCH2080KE MOSFET (1200V, 35A, 80mOhm)
    – 12V (or 15V) zenners (e.g. 1N5927B, 3W)
    – diodes IXYS DSEP 29-12A (1200V, 40nSec)

    – self wound chokes (1.5 or 2.5 mm2 wire) aiming for about 250uH each
    – capacitor tanking circuit bank made of MKP x2 300V 5.6uF

    Chokes and capacitors in mineral oil and this bath being cooled by a second water circuit
    Cooling pads of MOSFET and Triac also connected to this cooling

    Thank You

  20. Hey Martin

    This circuit does not work very well at input voltages above 80-100 VDC, if you want to build a off line supplied induction heater I suggest you look into circuits using oscillator ICs with feedback. This has entirely something to do with the internal construction of high voltage MOSFETs to do. Some people have made ZVS circuits that work from mains voltage, try to google your way to these, I do not have any experience with those.

    The triac voltage regulator is useless, it is only made for inductive loads like transformers and motors, not inverters presenting a heavy inductive load.

    The capacitors for the tank circuit will see a very high peak and RMS current, be sure that your capacitors can handle this or they will be the first thing to start popping.

    Oil cooling is not very efficient, direct water cooling isolated or floating is a better solution.

    Kind regards

  21. hello sir
    first of all im not an electrical student
    second I bought all components but the transformer
    I have a transformer(15v 5amps) which my friend bought for his project but he don’t want it so he is asking me to take it
    does the transformer(mentioned above) fits the circuit?
    if not what alternations can I do for the circuit so that it fits for that transformer if that is also not possible then what would u suggest me to do
    my project is based on melting

  22. Hey Mukesh

    15 V at 5 A is only 75VA, you need at-least 24 V and some good head room up to around 20 A, you can also aim for higher voltage, just remember to look up under considerations in order to find a transistor that can withstand the voltage you choose.

    Personally I have been using a 32V 20A transformer and not been able to melt metal yet, the power is simply too low. The frequency, coil geometry and metal you want to melt are all changing parameters in the formula for melting metal.

    I will try to find a 80VDC supply for my own, at at-least 20 A, hopefully I can start doing some melting then, it might also require putting more half bridges in parallel to handle the current and losses in switches.

    Kind regards

  23. Hey Selman

    As the supply voltage feeds both the control section and power section of this circuit, you only have this one voltage to break.

    If you only remove the supply from the gates of the MOSFETs, they will be in a uncertain state and will likely explode. If you only remove the supply from the power section the MOSFETs might keep oscillating from capacitively coupled energy through the gate junction, but then again, you have to use a large breaker and might as well just switch off the whole circuit.

    Kind regards

  24. Hello Mads,

    I an doing this for my school project.

    Can you please explain how you get 4,33 uF using 9 x 3 in series of 1uF?
    If my math is correct, the total capacitance of your 9 parallel of 3 in series would be 3uF, please correct me if Im wrong. The total capacitance of one of your series should be 1/3 uF and if we multiple that by 9 we get 3uF.

    Best regards,

  25. Hey Nikola

    You are right, where I wrote 4,33uF should be 3uF. I forgot to correct the capacitance when I did not have enough capacitors to have 13 parallel strings. Thanks for pointing it out, I will update the article and schematics asap.

    Kind regards

  26. Hey Mats,

    Thank you for your reply.

    Using a 3uF instead of 4,33uF would raise the resonant frequency to 133kHz.
    I plan to use capacitance of 4,4uF, that should give me a frequency of 109kHz.
    Do you think there will be any difference between my 109kHz and your current frequency of 133kHz?
    My heater should work fine with 4,4uF (thats 0,22uF 275V, 20 capacitors in parallel) ?

    Best regards,

  27. Hey Nikola

    A lower frequency will lessen the switching losses in the transistors. Depending on frequency you will also see different heating effects in different metals. Though a small difference of 24kHz is not going to do any difference in regard to replicating this circuit.

    Remember that the inverter voltage is Pi times the input voltage, so be sure that the tank capacitor set up can withstand that voltage.

    Kind regards

  28. hello sir,
    Im not an electrical student
    I bought all the things but the capacitor that u used for this project
    the capacitor I got is 80vdc 10000uf
    does it works ?

  29. Hey Muthu

    The large electrolytic capacitors are used to smooth the low voltage DC and to give a large reservoir of energy to avoid voltage sag that could potentially destroy the oscillator.

    If you are referring to the capacitor C1 in the schematics that is made up from many smaller capacitors, you can not use a 80VDC 10000uF capacitor, that capacitor sounds like a electrolytic capacitor which is not made for switching high frequency alternating current. The capacitance is also too high as it forms a LC circuit with the working coil.

    You need a capacitor as described under considerations.

    Kind regards

  30. no sir, not the c1 I’m mentioning, the large capacitors(white one with 70v 10000nf, sorry if im wrong but that is what I see if I zoom the pic 1 ) you used near the rectifier. for c1 im using 1mf 300v caps which I found in shop nearby at that time

  31. Hey Muthu

    Yes, you can always use a capacitor with higher voltage rating! 10000uF should be enough for the voltage not to sag too much.

    Kind regards

  32. I tried my best connecting all of the components but no use. every now and then something is burning, first my caps (all 27*1uf 270 v ) burnt then I replaced them with new, now diodes, sometimes my mosfets, I built a transformer with 32v and 25amps out. is that monster firing all my components?
    can I use c1 with 4uf 270 caps? 3 in parallel and 3 in series all are mkp type, and I have nomore time to wait to get the exact cap 1uf 270v to order online only 4uf 270v is handy

  33. Hey Muthu

    Yes you can use other values for C1, just try to keep it around 2 to 4 uF, but be aware of the high RMS current passing through them, they are likely dying to overheating.

    If you are using 32VAC, the transistors and capacitors will see 140VDC, be sure they can withstand that.

    Are you turning it on with a switch or through a variac? If the voltage goes below 10VDC, there is a big risk that both MOSFETs turn on and short circuit.

    Kind regards

  34. sir,
    do u mean switching the mosfet with switch or variac? no i dont have any of those, i just built a transformer with 32vac and 25amps o/p. o/p of transformer is connected to a bridge rectifier(35amps) with 4*80v 10000uf caps then straight to the terminals of the circuit. i exactly used the same components that u mentioned, now, i m using 4uf 270v caps instead c1 that u mentioned, thats it. i didnt try with 4uf 270v caps i ll tell ya after my experiment.

  35. venkataraman sahoo

    Thanks for your reply. Actually i want to know the length of coiled section (the region in which the material to be heated i.e. length of five turns of coil)

  36. This coil was made around a piece of 40 mm diameter pipe. It is 8 mm copper tubing and somewhere between 5 to 8 mm between turns, so I would guess around 70 to 80 mm.

  37. venkataraman sahoo

    Sir can i use capacitors of 1microF,400V instead of 1microF,275V as 275V is not available here

  38. Hey venkataraman sahoo

    Yes you can, I assume you mean the same kind of MKP capacitor just at a higher voltage rating. How good is your basic electronics knowledge? Using components at higher ratings only means that they might survive longer in the same circuit and maybe they cost more.

    Kind regards

  39. Hey Mads,

    Thank for your help. The heater is working flawlessly.
    I would like to make my heater less sensitive to the current.
    Can I use IRFP260N instead of IRFP250N?
    It has a higher drain current of 50A instead of 30A.

    Will the RDS(on) resistance of 0.04Ω make any differece?
    The IRFP250N has RDS(on) = 0.075Ω.

    Kind regards,

  40. Hey Nikola

    There is only one thing to do, try it. Not all MOSFETs can be driven in a self oscillating circuit like this. Especially higher voltage MOSFETs will not work, so you might have luck with one rated at the same voltage but higher current.

    Kind regards

  41. Hey Mads,

    Ive tried IRFP260 MOSFETs and they work great.

    I was wondering if you can send me some theory behind this circuit?
    I need to write about how this oscilator works and stuff.
    It would help me a lot.

    Looking forward to your answer.

    Best regards,

  42. Hey Mads,

    Can you please explain what is the role of these inductors?

    I cant find anything about them.

    Best regards,

  43. Hey Nikola

    If you search for DC choke you will find out that it works as a current source that supplies current in the off period of its side. A DC choke will ensure less ripple in the DC voltage at large loads. Just be careful not to use too small inductors.

    Kind regards

  44. You doing wrong math for finding resonant frequency calculated that involve measure of working coil and capacitor I tell you, but that is pretty lower freq which that is good. Always use AMP meter, just in case if you saw that number go over 10 amp fast. Clampmeter is expesive but it is good choose. It saved my 120 dollar project that time it jump to over 30 amp.

  45. Good day Mr. Mads Barnkob!

    I am so happy that I will be working on your project. I want to make one. I just had a problem with my power supply. I used a DC power supply. I measured it to be 18V. When I connected it to my circuit, it dropped to 5V. gosh!
    I am humbly asking for your help and expertise on this matter.
    I am planning to buy this 200Watts transformer with the following specs:
    Type: Auto
    Input: 220VAC,6Hz
    Ouput: 110VAC, 60Hz
    Capacity: 200VA
    Is this enough to run the whole circuit and make a good output?
    Thank you sir. I am hoping for your immediate response.

  46. Hello Abraham

    Your voltage dropped from 18V to 5V because the transformer was too small and not able to supply the current needed in this circuit, so the voltage drops dangerous low.

    The power rating of the transformer is too small and the output voltage is too high. You need a transformer with a maximum output of 40VAC and at best over 500VA, better at 1000VA.

    Read the chapter called considerations in the start of the article.

    Kind regards

  47. Noted.
    Thank you for that concern.

    I would like to share with you what would make my induction heater :

    I have the following electronic items:

    * 2 – BYV26E (or any fast recovery diodes)
    * 2 – Zener Diodes (10V, 1/2W)
    * 2 -470k ohms resistor (5W)
    * 2 – 10k ohms resistor
    * 1 – voltage regulator
    * 2 – IRF540 N-Channel MOSFET
    * 2 – inductor choke coil ( 0.12mH, connected in series)
    * 1 – 0.47uF MKP Capacitor
    * work coil

    Will I make sense if my induction heater will be made out of these materials??? (How will I show to you my circuit diagram? I can’t upload it here in your site sir.)

    By the way, I feel great and honored by your wise and quick response to my needs.

    Thank you so much sir.

  48. Hi Abraham

    I would advise you on using 12V Zener diodes, below 10V you might get in problems with the oscillator stopping with one switch open and the MOSFET will then die.

    Put those poor little TO-220 MOSFETs on a CPU heat sink with a fan, copper cored and just a tiny bit of cooling paste to get good contact, you will need some good heat dissipation to avoid burning them down.

    Other than that your materials list look fine 🙂

    I am happy I can help others have fun with the electronics hobby.

    Kind regards

  49. Greetings!

    May I know your skype account name, sir? Do you have one? I’d like to do my project online with your supervision. I’m a weakling when it comes to hardware. Thank you sir for your consideration.

  50. Hey Abraham

    I only do support in the comments. My time is limited and so that people with the same problems can find it via search engines.

    Kind regards

  51. Greetings!

    Hello sir! I am deeply troubled about my diodes. In replacement of that RURP880 diode, I am using a RGP15G diode with trr= 150ns . will it be okay?

  52. Hi Abraham

    The RGP15G diode is 15 times slower than the RURP880, it is however not slower than 5% rule of its Trr speed corresponds to 330kHz.

    You will have to try them out, maybe they are not fast enough to secure a stable self oscillation of the Royer circuit.

    Kind regards

  53. In the picture, what are those to-220 devices you have connected to the drains on the mosfet’s?


  54. Okay, so I decided to rebuild this circuit a bit nicer than I did last time and see if it had the same problem. Unfortunately, the MOSFET’s are still shorting out. It only happens when I hook up my 24V 50A supply, on my 12V 50A supply it does just fine. After about 3 seconds on 24V it will blow. up. Until that point it pulls about 30A, whereas when running off the 12V supply it pulls about 18A, then starts to drop down to about 10A as the MOSFETs get hotter. I am using all the exact same components that are in your schematic except the inductors and the switching diodes. For the diodes I am using MUR8100EG’s as they were the closest I could find. For the inductors they are 15A 100uH. I also took some scope plots of the gate voltage and the tank voltage if you would like to see them. Do you have any ideas as to why this would be happening?

    Thank you!

  55. Great thread, about diy induction heaters, everyone seems to be looking for more power and heat, I want to go the other way, and make a smaller unit as I’m only looking to anneal just the top 10 -15mm of brass cartridge cases.
    Keep up the good work.

  56. Hey Mads,

    I’ve built your circuit and it worked great although I only used an ATX supply’s 12V and it was sensitive to tripping itself.

    I’ve rewound a large transformer, currently rectified it gives 56VDC, though I could take some windings off to lower the voltage.

    What effects would higher voltage/lower current Vs. lower voltage/higher current have on the circuit?


  57. Hey Simon

    There is nothing wrong with build a smaller version of this circuit that operates at a higher frequency with much faster switches.

    Kind regards

  58. Hey Mike

    The Royer oscillator is a power hungry little circuit, it will eat almost all the power you can supply, so my experience is that with higher input voltage the current rises with it. You will almost square the power input with voltage rise. You can not just feed it low voltage and expect it to draw a high current.

    56VDC is as high as you should go with 200V rated switches, what is the voltage pulled down to under load?

    Kind regards

  59. Hey Mads,

    I just got around to hooking up a new 27000uF capacitor and well, things went worse than I was expecting. First round it blew an 8A fuse attached to the transformer pretty quickly. Second round I went fuseless, pre-charged the capacitor to 5V to lower the in-rush current and when I turned it on the transformer (microwave, cut the EI laminations to remove the secondary) vibrated violently loudly and dimmed all the lights in the house quite a lot. I killed it immediately, getting no voltage reading, and ran no further tests.

    Seems like 30V would be a lot more realistic.

  60. Hey Mike

    It sounds like there is something else wrong, even thought the inrush current into 27000uF can be larger, there is no way you should experience dimming lights, for that to happen the circuit should draw almost close to what your mains supply can handle just before the fuses pop.

    I think you should look for a short circuit or wrong polarity.

    Kind regards

  61. nayan kathrecha

    hi mads sir
    i am use a veriac in power supply for 40v to 60v ??????
    and i have veriac is 4 amp
    thanks in advance

  62. Hey nayan kathrecha

    You need a supply able to deliver up to 800 to 1000 Watt, so that is atleast 20 Ampere at 40VDC or 15A at 60VDC.

    Kind regards

  63. I’ve built an induction heater following the design that Marco posted, it worked first time and heated a rifle cartridge case cherry red in only 20 seconds. I turned the circuit off by unplugging the transformer at the mains, as other comments on the forum suggested the mosfets could latch up and pop if the low voltage supply was just turned off.
    It heated several cases ok, and then something went pop and the 25A fuse protecting the transformer output blew.
    The water cooling works, the copper coil and mosfets are cool to the touch, there is a fan blowing across the capacitor bank.
    I’ve changed the mosfets but still have a short somewhere. It could be one of the capacitors.
    More investigation needed.

  64. Hey Simon

    That certainly sounds like you have a short circuit when a 25A breaker trips, if newly changed MOSFETs still short out, it could be that one of them is always turned on, check the resistors and zener diodes around the MOSFET gates.

    Did you have each part out and measure if it was still okay? If you measure on components that are in the circuit you can get all kind of values that has nothing to do with the component you are measuring on.

    Kind regards

  65. Hi Mads,
    I stripped the zeners etc and checked everything again and changed one of the MOSFETS that was reading low ohms and circuit worked ok.
    The supply voltage is a steady 30V and current limited by fuse to 25A max, the voltage across the work coil was measured at 85 volts.
    I used a laser temperature probe to monitor all the components as the circuit was running and the workpiece was heating up. The Mosfets were around 35°C, the Zeners 50°C and some of the little .22uf capacitors in the tank reached 100°C, but most read 60°C.
    Then as before there was a flash/spark, the fuse blew and both Mosfets are dead.
    I’m not sure if the flash came from a Mosfet or not as there are no visible signs of failure on any of the components. The brass rifle case had only just begun to glow dull red, no where nearly as hot as previous tests.
    I’m using
    2 x IRFP250NPBF MOSFET, N, 200V, 30A
    2 x 1N5349B ZENER diodes 5W, 12V
    2 x BA159G fast diodes 1A
    16 x .22uf wima MKP Capacitors ( I tried to get .27uf but mega money)

    The toroids are T157-6 and measure 13.5uH when wound with 33 turns of 2.5mm section wire.
    Any suggestions as to why I’m getting failures whilst running for such a short period of time?

  66. Hi Mads,
    I forgot to say that the work coil is 6mm copper tube, with copper plates soldered to each leg that the mosfets heat sinks are mounted on and water cooled with a pump. Exactly as Marko did in his post ages ago that started this thread off.

  67. sanjay agarwal

    dear sir,

    earlier i had requested you to advise me on how to use an ordinary inverter
    ckt working in push-pull mode for induction heating for series resonant topoly.

    last time you were to take flight, i am still waiting.

    sanjay agarawal

  68. Hi,

    I have made this circuit and I want to use it as a miniature induction heater to heat a small wire under vacuum. Looks like I need to go to pretty high frequency and I want to go above 1MHz. At around 600kHz and above I get a lot of ripple in the waveforms and I suspect it due to poor switching of the mosfets. Any ideas?

  69. Hello Muzza

    You either need to drive the MOSFETs harder with a higher gate voltage or lower gate resistor to make it switch faster, but this also jeopardizes the health of the switches.

    I think you should try to find some faster MOSFETs and hopefully you do not need that high of a current rating for just heating small wires.

    Kind regards

  70. Hi Mads,

    Just an update for others: it looks like it wasn’t a switching problem after all. The issue appears to be more to do with stray inductance causing further (unexpected) resonances. In particular, stray inductance in connecting to the capacitor (bank). Not sure if this impacts on performance as much as it does on actually probing and making measurements.


  71. Hi, Mads!

    I already made an induction heater from here:
    and it worked fine.
    Then I tried this one and didn’t work, so I have a few questions.
    You said, that “the internal construction of MOSFETs with a higher voltage ratings makes them unsuitable for use in a self oscillating circuit like this Royer oscillator”.

    1. Are they unsuitable just because of the higher Rdson or are all higher voltage MOSFETs unsuitable, even ones with low ON-resistance?
    The thing is, that I have a 70 V power supply (Inverter welder – open circuit), so the IRFP250N-s are not good. However, I have 400 V, 26 A, 130 mOhm N-mosfets and they work on the version in the link. Should they work here?

    2. Can I just tap the work coil instead of using inductors L1 and L2, just to eliminate one potential problem (I don’t have LCR meter available). Would it work?

    3. Can you point me to some equations to calculate my own L1 and L2 inductors for a different work coil (for different inductance of a work coil)?

    I am a mechanical engineer and this is a project at university, so I would really appreciate your help,

  72. To correct my recent post:
    “However, I have 400 V, 26 A, 130 mOhm N-mosfets and they work on the version in the link.”
    I wanted to say, that they are good for oscilating, but that was at 10 volts (computer power supply under load). They didn’t work at 70 v.


  73. Hey Peter

    The two circuits are identical, only the center tapped work coil with single supply inductor vs. the work coil with two supply inductors is the difference.

    1: I have only built working versions of this using IRFP250N MOSFETs, as you can read in the comments others have used IGBTs with good luck too, but it is little differences between working and burning. Regarding suitable MOSFETs, it was explained to me as its a physical difference inside the MOSFETs gate construction on devices higher than 200V rating that makes them unsuitable for a royer oscillator like this, they simply need more precise and harder drive. You could try with a separate gate supply to avoid too high gate resistors, you could try using a high voltage rating zener diode, but be sure to choose one where you stay below MOSFET maximum gate voltage.

    2: You need the two inductors to use a single untapped work coil, which is by far the best in a IH. It will be easier for you to make water cooling with a single coil.

    3: The inductors does not need to be super precise, as you can see, mine are just regular machine tool wire wound around ferrite cores from a switch mode power supply. Here is some inductance calculators:

    Your second comment: You might have killed the gates of the MOSFETs from overvoltage when running as high as 70VDC. Consider using a separate supply as stated above.

    Kind regards

  74. Hi Mads!

    Thank you for your answer.
    I forgot to mention, I did supply signal circuit at 12V in the version without zeners (because of max gates voltage), just the work coil was on 70V. And grounds were connected.
    And yes, working coil and heatsink of MOSFETs is water cooled.

    So you suggest, to put for example 18 V zeners, 20 V supply for signal circuit and 100 Ohm resistors, instead of 470?
    Just thinking, maybe gate drivers would also help…

    One more thing, with 3. question, I meant, how can I get a ratio between L1 inductor and working coil. For example, this working coil has probably about 0,5 uH, L1 and L2 have 130 uH. What inductance should be L1 and L2, if I use for example 0,1 uH or 5 uH work coil? Is this linear or how can I get L1 and L2? If there are any equations at all for this example…

    Have a nice day,

  75. Dear me
    Can we modified the CLF (compact light florescent)circuit into induction heater circuit thank you

  76. Hey!

    I’ve been experimenting a bit with different induction heaters to anneal rifle brass (also mentioned by someone else above here). I’ve had problems with my work coil being too small. It needs to have a small amount of windings (2-4) since it’s only the upper 10-12 mm of the brass that must be heated. Do you think your driver is able to drive such a small coil?

  77. Hey Trond

    A smaller work coil will result in a increased resonant frequency from the lower inductance, this circuit will struggle to switch properly at too higher frequencies.

    Exactly how high it can go I can not say, but you could increase the tank capacitor capacitance to still have a resonant frequency around 100 kHz. But try to keep it at maximum 5 uF.

    Kind regards

  78. Thomas Ursini

    What about driving the gates with a 12v laptop power supply.But then wher would you put the little negative wire.Duh.

  79. Hey Thomas Ursini

    Connect the positive gate supply before the current limiting resistor, not just the gate resistors. Connect the negative rail to the existing negative rail and put a 1 uF capacitor between negative and positive gate supply.

    Kind regards

  80. Thomas Ursini

    Thanks mads,i found an led driver that outputs 12v and 1.5a.And when you say to connect the positive supply of my gate driver to the current limiting resistor and not just the gate resistors I didn’t realize there were more than two resistors befere the gates. And one more thing sir I have 35v × 35a power supply with 250v mosfets.However the vgs is still +or – 20 $o do you thing I should lower my voltage to 30 which would up my amps a little because of the number of primary turns onmy transformer.

  81. Hey Thomas

    This build and schematic does not have individual gate resistors, I have recently used that when I paralleled more MOSFETs. If your supply voltage is not higher than 35VDC, there is no reason to use another supply for the gates, perhaps a higher value for R3 and R4.

    Vgate-source rating of +/- 20V is normal for almost all MOSFETs and have nothing to do with your drain-source voltage. The current limiting resistors R3 and R4 and the zener diodes D1 and D2 ensures that the gates are not subject to overvoltage.

    Kind regards

  82. In the attached picture it appears that you do not have your cooling system loop attached since the end of the copper tube is bent back on it’s self and appears to not be connected to a cooling loop. Is this correct or do you have an extra piece of copper attached to the tube for electrical connection points and the main tube line continues to your cooling pump? It is hard to tell because one end of the tube is obscured by the capacitor array and the bend on the top of the work coil is in shadow. Have you tried a dielectric fluid for cooling? it seems to be the fluid of choice when cooling electrical components since it is non conductive. Which in the event of a cooling leak you would not experience any component damage or shorting.

  83. Hi Bryan Faris

    You are correct that there is no cooling system connected. I only did those few filmed test runs and even that was actually enough to damage the work coil from excessive heating. The copper got somewhat miscoloured.

    I use distilled water for water cooling, it is a real bad conductor and there are no significant losses in a meter of hose. This can be used on a hobby level as it is not build to run industrial production 24/7, the little corrosion or electrolysis that will happen is too little damage to the coil and pipe compared to it being a test setup.

    I would properly use distilled water with anti corrosion additives in a more permanent setup, id rather have hot water that can short circuit components and burn me slightly, compared to protecting components by using oil, that on the other hand is no joke to get burns from.

    Kind regards

  84. copy that and copy that do you use some sort of heat exchange or a small radiator to send the output flow through for cooling or just dump it in a five gallon bucket with ice in it?

  85. I just circulate the water in a bucket, I have not yet cared to freeze down distilled water to add ice. But even a bucket of water takes a while to heat up above 50 degrees Celcius where I will stop, but that is only because the pump is not made for hot water.

    Coil would not care for boiling water, but the capacitors are only rated at 105 degrees Celcius, air bubble forming from boiling could possible be a issue in the tubing.

  86. Hey Jahsus

    If you want a upgraded version of the Royer/Mazilli ZVS circuit, you will have to find one that uses active gate drive circuitry. Besides that there is little difference between the different versions of the original schematic, differences there is down to components changed to suit the MOSFET or IGBT used.

    Kind regards

  87. Hi Mads,
    What purpose do mosfets T1 and T2 (or your Q1 and Q2) in this circuit serve? And could the capacitor in my schemeatic replace D3 and D4 in your schematic?
    Thank in advance,

  88. Hey Mike

    The MOSFET are the switches that are alternating turning on and off the current that flows through the primary coil L1.

    The capacitor C1 on both schematics is the resonant capacitor with the primary coil.

    D1 and D2 in your schematic is D3 and D4 in my schematic. The difference is that I use fast diodes for the power section and zener diodes to clamp the gate voltage, your schematic only relies on a zener diode for clamping gate voltage and oscillation. The zener diode is however not fast enough if you ask me.

    Kind regards

  89. Hi,

    A few of us have been working on this for a project.

    Our components closely match yours, except for our capacitor bank which provides 3uF when disconnected and 0.2uF when connected, our coil is showing 0.432uH on our LC meter, but when operating, we are getting a frequency of 5 mHz which is 40 times higher than the theoretical value.

    1/(2pi*sqrt(3E^-6 x 0.432E^-6)) = 133 kHz

    The result has been a lot of blown mosfets and more recently, bridge rectifiers.

    If you could provide some guidance as to why the frequency is so far off, that would be great.

    Currently, our capacitor bank isn’t connected directly to the coil as shown here. Our voltage is 23VDC smoothed by a 20,000 uH capacitor. Our mosfets are above specs and bridge rectifier is 50A. Our transformer can only supply 100 watts.

    Thanks for any suggestions you can make.

  90. Hey Rocket

    You can not measure the capacitance when the bank is connected to the coil, you get the self capacitance of the work coil in parallel with the bank and it gives a whole other reading.

    Which capacitors did you use? How many in series / parallel?

    A 100 Watt supply is not enough for this circuit, it will either pull the voltage down due to the load or get stuck in over-current protection if it is a SMPS.

    Kind regards

  91. Hi Mads,

    What effect does the voltage have on
    A) The switching times … can’t find anything about it in the data sheets?
    B) How do the switching times of the push / pull topography effect the tank caps charge rate?
    C) Does the voltage have any effects on the freq? (I thought it was just the LC component of solenoid and tank?)

    Thanks in advance!

  92. Hello Jahsus

    A) Gate voltages from 4 to 9 VDC gate-source has a massive impact on switching characteristics. See the datasheets for ”typical output characteristics” and “typical transfer characteristics”. This is however not something you should worry about as you should not supply it with less than 12VDC. Higher voltage will make the MOSFET switch faster, but MOSFETs does not tolerate more than +/- 20VDC gate-source.

    B) The slower your switches go from off to on, the more energy is lost in the switch, this energy loss will not make it to the tank capacitor.

    C) The resonant frequency is formed from the LC product, voltage does not matter.

    Kind regards

  93. Thanks Mads you are a true gentleman.

    I do not understand the +/- 20V gate-source part. I have been reading this:

    …and from it I have got, the GATE is pretty much the same as the BASE of a normal NPN trans. Now the Vbe for silicon is 0.7V and it will turn on when it is 0.7V above the Ve.

    So … the GATE voltage needs to be (for an IRFP260N) according to the data sheet 2V – 4V above the source voltage (10 – 40V in the schematic) ?

    The article then goes on to explain;

    ” Delivering a higher voltage (up to 12v) will not damage the device or cause more gate current to flow but supplying a minimum voltage will alter the current capability enormously. ”

    But in your circuit you can use upto 40V and have said in the comments as high as 60V… yet the article says UP TO 12V … this just leaves me confused especially when it says;

    “Whenever the Gate voltage exceeds the Source voltage by at least the Gate Threshold Voltage the MOSFET conducts. The higher the voltage, the more the Mosfet can conduct.” … and you would want it to conduct better hence the higher voltage.

    I understand higher V increases the switching speed, which in turn causes less heat.

    I do hope I am not annoying you…….

  94. EDIT: … and you would want it to conduct better hence the higher voltage, but you don’t want it to go too high … so where is the trade off?

  95. Hi Jahsus

    You seem to understand what is going on and find the right information, you are doing good.

    The gate voltage is limited by the 12V zener diodes D1 and D2.

    You will not have more gate current flowing, but higher voltage at the same current delivers more energy over time. The gate capacitance has a certain energy needed to switch and it is this we are sourcing current into.

    I am happy about answering questions, it also helps me when I at times have to read up on stuff and it keeps me in the loop 🙂 We need more electronics amateurs and we only get that by helping each other.

    Kind regards

  96. Hi Mads,

    I can’t seem to find anything about why the mosfets and the tank circuit see Pi (3.14) times the supply voltage. Any reading material on this?

    Thanks again!

  97. Hi Jahsus

    The mazilli driver is basically two single ended Class E oscillators mirrored around the same work load. It is the inherent switching mode of single ended class E that makes the switch see a voltage 3-6 times higher than the input voltage. It depends on how much the resistive load moves outside of the driving point the oscillator was designed for.

    Kind regards

  98. Hi Mads!

    ” It depends on how much the resistive load moves outside of the driving point the oscillator was designed for. ”

    Surely we could account for this by keeping the transducer diameter within an acceptable size? This wouldn’t allow for oversized loads to be used in the coil limiting the current draw? Am I correct in thinking this?


  99. Edit:

    “This wouldn’t allow for oversized loads to be used in the coil *thus* limiting the current draw?”

  100. Hi Jahsus

    I guess that Rload to the switch in a Class E oscillator would be the load presented by the work coil. I do not know how much the resistive load changes with the size of the inserted work piece.

    Richie Burnett covers some of the Class E here: and mentions the double ended class E at the bottom, which is essentially the royer induction heater. You would have to do some more research yourself from how Class E works, the voltage envelope, voltage profiles, how different load affects the Q of the system etc. There is a lot to read and I do not have the time for that right now 🙂

    Kind regards

  101. hello sir, thanks for your good work. am constructing 15kw induction furnace, running frequency of 1khz, with15kw(15000w) transformer so I can melt steel, like 20kg of steel. the 15kw transformer input at 220vac while the output at 180vdc which will be supplied to irfp260n mosfet arranged in full-bridge using ir2110 mosfet driver. since irf260n is 300watts 200v and 49a, I intend to use a whooping number of 100 pieces to get 15kw, am using SG3524 to get my 1khz frequency by altering its value of RT and CT. my real problem now is, how many turns should I wind the working coil around 8inch diameter crucible to match 180VDC @1khz and what value of capacitor to attach in parallel on the working coil, how many amp diode to use. you
    thanks so much as am looking forward to hearing from you soonest.
    please you can send guidelines or detail to my e-mail address:–


  102. Hi Ikenna

    Your real problem is not calculating the frequency of a LC circuit like the work coil and resonant capacitor. Scaling this circuit up to 15kW through 100 parallel switches is a complex and time consuming task that will require careful component layout.

    1 kHz switching at 15kW will be near impossible to be near from the high power switching in the audible range.

    15kW is also nowhere near enough power to melt 20 kg of steel.

    You simply do not know enough about what you are building, you want to build something too big to start with before you know the basis of induction heating.

    You need to move on to PLL/microcontroller controlled induction heater using large IGBT bricks. Melting 20 kg of steel would require much more power than you have available in a normal house hold.

    Kind regards

  103. Hi Mads, thanks immensely for your respond. is there any other way I can beef up the inverter to 15kw without using caches of mosfets, because to the best of my knowledge, number of power mosfet plays a vital role for power output of inverter, that was why I want to use 15kw transformer, with output of 180VDC. I didn’t get what you meant by 1khz @15kw is near impossible, please can you shed more light on that.
    if 1khz @15kw can operation efficiently, like how many KG of steel can it melt.
    I want to build low frequency induction furnace, because from the finding of my studies, I discovered that low frequency inflict deep heat penetration on metals, that’s why I choose 1khz frequency for myself.
    from your statement, it seems my set up is out of ordinary, please can you give me some detail on induction furnace.
    I have little knowledge about microcontrollers, though I have studied Arduino. currently I have “Arduino Uno” but am not yet versed in C/C++.
    please can you give me more info on induction furnace. thanks
    best regard.

  104. Hi ikenna

    You should look into PLL controlled induction heaters using large IGBT bricks. The Royer induction heater is a much simpler design and I won’t recommend trying to upscale it as the design has several severe flaws that often makes it blow up the MOSFETs.

    The reason for me to advice against 1 kHz switching frequency is the noise level will be unbearable as it is in the audible range hear-able by humans.

    Kind regards

  105. hello mads, thanks. please can you give me formular for calculating resonance capacitance and its frequency, and calculation for workcoil.
    do you mean that voltage supply to mosfet source-drain should not exceed 60vdc. thanks

  106. Hi Mads , first I want to wish you a happy new year and many others new projects !
    I want to use your induction heater to heat some aluminium rings at about 70-80 C degrees in approx 2-3 mins . The diameter of rings is 38 cm , the section is 2 x 2 cm and they have approx 200 grams . Do you think it is possible ? What kind of coil should i use ,a normal coil (with a diameter of 38 cm) or a pancake coil (with a medium diameter of 37 cm ) ? What kind of wire should i use for the coil and how many turns ? Should I made some modification to the scheme ? Than kyou very much for your support !

  107. Hi,Mads Barnkob.
    I am a student of an electrical engineer.
    i want to build an induction heater,so i want to know how can i calculate the length of working coil,number of turns and material of coil.
    waiting for your reply.

  108. Hi Titi Duca

    Your biggest issue is properly the large inductance such a big work coil would result in, the resonant frequency of the circuit would maybe be very low, resistance higher and I am afraid this circuit is not powerful enough to use such a large coil and warm such a large work piece as your aluminium rings.

    Honestly I think you would have more luck with a large old cooking plate since you do not need it warmer than sub 100 degrees Celsius.

    Kind regards

  109. Hi Mads

    Would I need to make any changes to the circuit if I only want to raise the work piece to about 120 C for about 20 min continuous?

    Kind Regards

  110. Hey Trenton

    There is no control input for this self oscillating circuit. For your task you would need to add a temperature sensor and set up some simple on/off control of the power to the induction heater.

    But you will not get any precise and stable temperature like this. You would need a proper induction heater with a driver where you could vary power output from a PID controller to keep the temperature stable.

    Kind regards

  111. Torgeir Fredriksen

    You have to cool it by mounting it on a proper heatsink. Read what Mads wrote in the conclusion:

    “A good and reliable oscillator as long as supply voltage is kept within safe area of operation for the MOSFETs and only short run times are used unless there is used good components and water cooling on work coil, MOSFETs and capacitors.”

  112. Hi Abiodun

    You have to explain your circuit and operating parameters before we can help you in any way. We got no idea what you made, which components, supply voltage and so on. For the best possible advise, also upload a picture of your setup.

    Kind regards

  113. Hi Mads me again,

    Few questions,

    Firstly, from Richies site I can’t seem to work out if the schematic is LCLR topology or not with L1 and L2 parallel inductors ( I seem to remember you said they acted as DC chokes to counter DC ripple voltage.) Or would it need seperate inductor acting as the Lm impedance matching component. If so how would I be able to work out the reactance Xc is it just a case of 1/ 2pi’fC where F would be 100kHz and C would be the tank cap of 3uF? I am still a bit confused about the power side of things.

    Thanks again.

  114. Hi Jahsus

    Inductors are indeed voltage storages and that is also their primary function in this circuit. For your example, Xc would be 0.5 Ohm.

    Shorting the work coil will be the same as having a work coil with fewer turns, this does two things, the resonant frequency goes up, depending on how many turns you short out, it might get into MHz region. A higher resonant frequency will most likely result in MOSFETs not being able to switch that fast and they will dissipate a lot of heat or it could simply ruin their self oscillation and explode 🙂

    Kind regards

  115. Torgeir Fredriksen

    I just received the chinese version bought from ebay 🙂 Switching is done by 2x IRFP260N and the MMC is 6x “BM” 0.33uF 630/122V MKPH capacitors in parallell. Quite impressed by the transistors but not sure about these “BM” branded capacitors. And they are only half the capacity you suggested to avoid disappointing results 🙂

    However, the provided coil is about 7 turns 50x50mm (diameterxlength) and seems to be made of automotive copper brake line. First tests shows good performance but the item being heated has to be very small to avoid high currents. If not monitorig current I can imagine this circuit with the provided coil breaking down very fast…

  116. Hey Torgeir

    The MMC is properly fine, a total of 2 uF is good enough for this circuit. As long as its MKP, maybe X2, capacitors used in this circuit, there is no real difference between their performance or expected life time. Real induction heater capacitors should be used or some large GTO snubber capacitors with a high Arms rating.

    I am sure these are build from the cheapest available components, hence the brake line copper tubing.

    Kind regards

  117. Torgeir Fredriksen

    Hi again Mads,

    As mentioned previously I have started off with the chinese made version which is almost identical to yours. I want to improve it a bit though, cause as you said it’s made to be as cheap as possible.

    The MOSFETs are ok, 2x IRFP260N. Resistors and diodes as well. Bank capacitors are made of 6x 0.33uF MKP “BM” for induction heaters. I want to increase the capacity a bit so I have ordered a few more of these. Have also experimented with various work coils and the performance is actually not bad.

    What concerns me a bit is that the drossels get really warm, so I have to do some changes to those. They are made of toroids (yellow core), about 25-30 turns of 1mm2 wire. Outer diameter is 25mm. Have not measured the inductance yet, but I guess it is within range. They get hot even if there is no load (material inside work coil).

    I know this question should be addressed to the manufacturer, but I believe you have even more knowledge of how this circuit works. Do you think rewinding the coils with a thicker wire would help? That is my plan (as well as increasing tank capacity) but I would love your opinion on it. Thank you 🙂

  118. Hi Torgeir

    If you add more capacitance in parallel, you will lower the resonant frequency, have better RMS current handling for at the MMC and the lower frequency might also result in less losses in the iron powder cores, iron powder cores, as energy storage inductors, starts to decline rapidly in function after 200 kHz, after this MnZn cores should be used.

    The iron powder cores are used because they are cheap, its only powder and glue put together under great pressure, ferrite coes needs to undergo a sintering process which is more costly.

    A thick wire on the core will not help, it is losses in the core itself and not from the wire. From room temperature to 85 degrees, there is an increase in approximately 15% in permeability in cores and they will completely collapse above their curie temperature at 155 degrees, so watch out for the temperature.

    The changed frequency might also influence how your work piece is heated, different materials heat differently at different frequencies, I got a paper on it that I will upload to the file archive, see bottom of menu, tonight.

    Kind regards

  119. Torgeir Fredriksen

    Thank you very much for your feedback, Mads. I really appreciate it. While waiting for the caps I tried replacing the 2 coils with a couple of ferrite cores I had, which I wounded about 9 turns of 1.5mm2 wire around to make it about 100uH which the original coils were.The new coils did not get as hot as the original ones so I believe what you said above makes sense. They did however get hot eventually, after a few minutes (even with a fan above them). The heat was indeed induced by the core material and not by the wire as you said. Not sure about the quality/material of these cores, they probably came with some computer cabinets I have previously built.

    Is it normal that these coils get warm and should they be cooled as well? Or should I find some quality MnZn cores instead? Still waiting for caps to increase the tank capacity, maybe that would eliminate the problem?

  120. Hi Torgeir

    You could use cores with a larger cross section area, just keep an eye on inductance so that you do not begin to limit the current inflow too much. It is normal that inductors in variable frequency drives are built so that mounting against heat sinks is possible as they do get warm at high power levels. We are abusing the components and that is also why you do not see this simple design used in commercial units, it will not last over time 🙂

    Kind regards

  121. Torgeir Fredriksen

    Mads, when I increased the bank capacitance from 2uF to 4uF the idle current was doubled as well (@36V it went from 4A to 8A using the same work coil). I tried making a work coil half the inductanse (from 1.4uH to .7uH) to keep the resonance frequency the same, but then the idle current got even higher. The work coil gets hot from 0-100 in about 5-10 seconds. Does that make sense or…?

    PS! Waiting for the frequency paper as well 🙂

  122. Hi Torgeir

    I have not done specific experiments on LC ratios, but one reason I can think of is that you do not only change the resonant frequency, you now also have a larger energy stored in the C part that now circulates in the LC circuit, which most likely is the reason for the rise in current.

    Kind regards

  123. Torgeir Fredriksen

    Thank you very much for your feedback, Mads.

    I’ve been testing this out for some time now with the main goal of being able to heat a mini crucible (40x40mm) without any components overheating. I know you told that this circuit design is only capable of short run times, but…

    I have used the chinese PCB (for simplicity) with 2x IRFP260, modified L1/L2, kept the 2uF bank caps but made a new work coil from 8mm copper tubing; 6 turns 90mm diameter to make sure the crucible isn’t too much of a load. At 49V current is about 10A and with water cooled work coil and a cheap 10mm 14W tunnel fan to cool the other components I kept it running for about 10 minutes without anything getting too hot 🙂

  124. Hi Torgeir

    As you have experienced, a high voltage and low current makes this run cooler, but that was purely unloaded(only crucible right?), as the work piece gets hot and eventually melts the load will go up. Depending on your power supply your voltage will either sag and current rise, you will still see more heating due to higher currents, but not as much as if your power supply is powerful enough to keep the voltage high and just supply more and more current, then you will likely see it burn 🙂

    Looking forward to see your results.

    Kind regards

  125. . There are a few $500, 1000watt, hand held ZVS heaters running on 120 VAC line voltage for freeing stuck nuts and bolts on cars. The interchangeable heater coils are just a few turns From what I can gather they just rectify and filter the line voltage and apply it straight to the ZVS so that means the filter caps have around 160 volts on them with no load. Portable induction cook tops run this way, off the line avoiding a big high current, low voltage power supply. Even switching ones get big and expensive at a Kw. I wonder if I could re pourpose one of them, though the coils seem to be several tines more inductance. The ZVS s on E Bay are all high current, low voltage. I recently read that ZVSs are less fussy running at higher volyage and lower current. I want something portable and cheap with a few turn work coil and don’t want to lug around a big power supply and I am sure not going to spend$500. I realize the hazards of not having AC isolation.

  126. Reading through your comments, I see you mentioned another type of power oscillator that is better adapted to running on a high voltage, lower current power supply. any more info? Thanks.

  127. Alexis Salmon

    Hola, tengo una pregunta. Yo estoy haciendo el mismo circuito pero no me funciona estoy utilizando un MOSFET’s IRFP064N en lugar de del IRFP250N y estoy utilizando una bateria de carro para alimentar el circuito. Y no funciona nose que pasa . todo lo demas es exactamente los mismos componentes. Ayuda porfavor

  128. Torgeir Fredriksen

    You need higher voltage to drive the circuit. A car battery is only 12V….

  129. Alexis Salmon

    Pero es que con la bateria de carro se quema un MOSFET instantaneamente en el momento que encendemos el swith y no sabemos porque pasa esto.. ayuda!

  130. Alexis Salmon

    y otra pregunta si habrá algún problema si remplazo la bobina de trabajo por una bobina plana como las cocinas de inducción?

  131. Alexis Salmon

    But it is that the battery car burns a MOSFET instantly when we turn on the swith and do not know why this happens .. help!

    and another question if there will be any problems if the work coil replacement by a flat coil and induction cookers?

  132. Hi Alexis

    As Torgeir points out, the battery voltage might be too low to drive the MOSFET out of the liniar region, where it dissipates a lot of energy and will burn up if its too much. What is the voltage on the battery when you apply the load of the induction heater? If you have access to two batteries, try to put them in series for 24 V.

    There should be no problems in using a flat coil with similar inductance as a normal helical work coil, this should work fine as a induction cooker. But try to keep it small in diameter to start with.

    Kind regards

  133. Hi Alexis

    Your battery is defective then, a 12 V lead battery should never get below 11 V. The induction heater circuit is likely to destroy itself if the supply voltage gets below 10 V, that is written in the article!

    Kind regards

  134. and I’m desperate for help. This is a picture of my circuit I’m controlling a microprocessor called arduino one to control pulses in lguar of the circuit of the diodes and resistors help 🙁

  135. Hi Alexis

    If you replaced the components that main the self-oscillating nature of the Royer oscillator, you have basically converted it to a double single switch converter.

    If you want to use a micro controller I suggest that you just control a regular half- og full-bridge instead of modifying a Royer oscillator. Kind of how my TL494 flyback driver works.

    Kind regards

  136. Hi kewelscience

    What input voltage are you using, is power usage the same and do you get the same results in term of heating power from both types of MOSFET?

    Kind regards

  137. Hello
    I would like to study your circuit but it is complicated for me.
    Would you mind explaining your circuit simply?
    Thank you

  138. Hi Nutchapol

    Search the internet for “Royer oscillator” and you will find many explanations on this kind of circuit.

    Kind regards

  139. sir,
    What will be the inductor coil value and the capacitor bank value ? In which way i can calculate that and how much magnetic flux create in the working coil .if you give me the calculation that will help me a lot.

  140. Hi Saikat

    For the best results, stay somewhere around 500 nF to 4 uF, work coil some 3 to 6 turns, try to keep the switching frequency below 100-200 kHz. It is calculated just as a regular resonant LC circuit.

    Kind regards

  141. sir,
    what will be the magnetic flux in the working coil ,how can i measure it mathematically .is there any equation by which i can measure the magnetic flux .

  142. Hi Saikat

    Find the calculations for magnetic flux in a solenoid and use the work piece as core. Either measure or use the results from my measurements when you need the current for the calculations.

    Kind regards

  143. If resonant frequency is 1000 khz for 12VDC supply ,60uh inductor and 630pf capacitor then what problems will happen.also want to know how much magnetic flux will create in the working coil .can you give me any mathematical equation for measuring the magnetic flux

  144. Hi Saikat

    I gave you some clear points on what to do, now do your research and learn something. I am not doing your design/home work/whatever for you.

    Kind regards

  145. Pierre-Alexandre


    Thanks for the article, it is really interesting and well explained.

    I have few question for you. I did an induction heater following your plan. I used the same MOSFET, fast diode, 12V Zener and the same resistances (1k and 470ohms). My choke inductor are 146 uH. Capacitor of the tank circuit is only 1uF but inductance of my coil is higher than yours (for a frequency of around 160 kHz). Capacitor and coil are water cooled). I added some 30A fuses and a switch at the power source (a double switch opening positive and negative).
    I tested the circuit with a tension source at 15V. The circuit worked. It oscillated at around 160 kHz and the current was around 5A.

    I tried the circuit at 36V using 3 car batteries in series. After few second I saw a flash near one of my MOSFET. I stopped the circuit. MOSFET and other parts looked fine. I then tried my circuit at 24V but one of the MOSFET exploded (but maybe it was damaged from the first test).

    I am trying to understand the cause of the flash. I don’t think the components where misplaced as it worked at 15V. Could it be an overvoltage? If yes, what could be the cause? Does changing the values of the components of the tank circuit change the voltage at the MOSFET (I don’t think so but I am not so familiar with this circuit). If I changed the values of R1 and R2 or R3 and R4, would it change the voltage at the MOSFET ?
    Do you have any ideas where the problem could come from ?

    I know you talked about where to place a switch in other comments and that depending where the switch is placed it could damage the circuit. Where is the best place to put a switch in the circuit?

    Sorry for my imperfect English, this is not my first language.

    Thank you


  146. Hi Pierre

    The voltage across the MOSFETs, that is dumped into the LC circuit of the tank capacitor and work coil is pi (3,14) times the input voltage, so the voltage across your MOSFETs was about 113 Volt. Remember that this is a very power hungry circuit that will take almost all the power it can get, if it was powered from a 500 VA transformer, the voltage would drop once the current was high enough to saturate the transformer and eventually that limits the power output of the circuit.

    A battery is able to supply hundreds of Ampere for short times when the load is very low impedance, as the induction heater LC circuit is. I think you pumped so much power through the MOSFETs that the die melted inside of it.

    It is best to place a switch so that it turns off the supply to the inductors that supply the LC power part and still have supply to the gates of the MOSFETs.

    Kind regards

  147. Hi,
    I’m very interested in your project and I’m as well building one for myself.
    Since I am waiting for my ordered transformer (which could do some high power thingy) to be delivered, I’m working on a circuit that yields less to try it out first. I only have a 24V-240W switching PSU, so I’m aiming to get a small circuit that works with 6-7A (no load) and see if it can oscillate.
    However, my MOSFETs tend to die almost immediately after I let the current flow through the circuit. After blowing up around six MOSFETs – well, they didn’t explode, though, only died and all leads got shorted), I decide to take a break and see for the reason.
    The following is my part list:
    -4700µ x4 electrolytic capacitors to maintain the PSU’s output
    -12V Zener, 10kOhm 2W resistors, FR107 as fast diode
    – 220Ohm 2W gate resistors
    – IRF540N (100V, 44mOhm, as the supply is 24V). I also used IRF640N (200V, 150mOhm) before but they didn’t work.
    – 0.91µF x 4 tank capacitors (I find through simulations that lower cap will yield less current, so I want to test that out. It might be the reason for my MOSFETs to die)
    – 2mm diameter wire for 5-turn, 5-cm-diameter coil. It is expected to be around 1-2µH, I can’t measure its inductance, though, as my RLCmeter can’t measure it.
    – 2x 1000µH toroidal coils. Through simulations, as well, more inductance will reduce the HF. I also have tested with 150~160µH coils.

    With simulations I got around 3-4A current. However, in all cases the PSU’s voltage just dropped and sparks appeared at the interrupter when I closed the circuit, enough to melt and stick the metal contacts together. Even though I tried to save the circuit by open the contact almost immediately after 1-2 seconds, the MOSFETs still died.

    Do you know what is the reason for the circuit failure?

  148. Also, I’m more curious about how to choose the correct MOSFET. Basically I choose a MOSFET with Vds max >= 4 times supply voltage, and Rds is as small as possible (and inferior to 150 mOhm). Are there also other criteria, like drain current and stuffs? If the MOSFETs malfunction in this circuit, which are usually the causes?
    As for the fast diode, what is the recommended voltage rating? Does it depend on power supply? The FR107 doesn’t seem too fast enough, but it is the only one with high voltage rating that I could find in my place. There is also the 1N4148 which is like fifty times faster, but it only has 100V rating. If my circuit uses low power supply, could the 1N4148 work?

  149. Hi afk

    I can think of two reasons that your MOSFETs fail.

    The first could be that the FR107 diode is underrated, maybe 1A is not enough despite the current limiting through the 470R resistors. It has a Trr of 500 ns and that is fast enough for a circuit that switches around 200 kHz.

    Second could be that the SMPS is not powerful enough, when you load it down heavily with a power hungry circuit like this, it will look like a short circuit to it and it will shut down. It will then repeatedly try to start up again and then shut down again. This might just get the voltage high enough for the circuit to start switching, but still be below 10 VDC where you can no longer be certain of which state each MOSFET is in and it might short circuit and die to too high die temperature.

    Kind regards

  150. Pierre-Alexandre


    I have another question.
    I was able to make the circuit work and to heat some metal (I built a tension source to replace the batteries and made few modifications).
    I still have some problem. The current in my circuit is low in comparison with your circuit. At 35 VDC (35V measured after the drop due to the load), I only have 10 Amps when the coil is loaded. I am trying to understand what might be the cause. It is not because of the choke inductor (I tested many value) or the source (it can provide more current than what I need). One of my hypothesis is the value of my tank circuit capacitor. Its value is only 1uF, which is lower then what people normally use for that circuit. I use a coil that has a higher inductance than you to have a frequency that is around 120kHz. Can that be the cause of the lower current in my circuit ? I thought value of the capacitor would only affect the frequency and that I could correct with the value of the inductor coil but could it affect the impedance of my circuit when oscillating ? Do you know anything else what could limit the current in my circuit ?
    I have a second question. What are the advantages of using two choke inductors over one that is connected at the center of the work coil. Is it only because it is more simple to build or it has other advantages ?

    Thank you for your help

    Have a nice day


  151. Hi
    I have practice this mini circuit. Here are some issues I would like to discuss wish you and some armatures.
    1. In my test,two inducers are made of 1.5m copper wire ,which bended by hand in a Mn-Zn core with the internal diameter 18mm ,outer diameter 35mm and height 15mm. These two inducers create 300uh each. Under 40v 50A DC power supply, my test works well(2mos are irfp4332,50A,300v ). Result in melting about 15g copper easily, maybe around 900-1200w. In your circuit, you emphasize these inducers must between 40-200uh size. Is it could be made larger? Maybe 500uh or even 1mh? Is the inducer the larger the better here in this circuit?
    2. Pierre-Alexandre in his last comment says, he want to enlarge the capacitors, I use a coil with about 7 turns, diameter 55mm, and the LCR meter shows the coil create 2uh induction itself. When I first try my capacitor under 1.5uf, the heating result is out of satisfaction. When I add my capacitor to 2.5uf, the expected frequency goes down, 80kz, and the result gets better than before. So, without considering hearing noise, could this capacitor added even larger aimed to increase hoting effect? Maybe 5uf or even 10uf ?
    3.In view of unstable ZVS circuit design itself. I want use TL494 to provide a stable signal source supply. See attached pictures1. Is this could create a 900w around power result?
    4.To enlarge the ZVS circuit power, I want use 4 mos, 4332. I develop a circuit like this. Is this possible to enlarge output power? attached pictures2.
    5.Two or three production are sold on ebay about this small ZVS heater. Some of them are made of China or England. Do you want to make some stable ZVS heater like that by your own. I am looking for your ZVS product.
    Have a good day
    Yours Leon.

  152. Hi Pierre-Alexandre

    Higher impedance in the LC circuit will limit the current that can flow in it, so lower the impedance of the work coil and maybe use a larger capacity for more circulating energy.

    The two chokes was used to avoid center tapping the work coil, so it is simply a question of making it easier to change between different size work coils and to water cool it.

    Kind regards

  153. Hi Leon
    1) A inductor is a voltage source and a larger inductance will make it able to supply that for longer, but it comes at a voltage drop cost.

    2) Heating effect is not only power dependant, but different materials also heat differently at different frequencies. Though I agree that there generally is better results with larger than 1-2 uF capacitance.

    3) You should look into the induction heater LCLR designs, that is basically what you have drawn a simplified version of.

    4) You can use more MOSFETs in parallel to be able to push more power through, “rogerinohio” on youtube/4hv once made one with 8 MOSFETs on each leg, you just need to make sure they are on the same heat sink, has individual gate resistors that are matched and good enough power supply for the drive voltage.

    5) I have no plans on trying to compete with the ultra cheap Chinese ZVS IH on ebay. If anything I would make a LCLR kit that would run stable. People want stable, not cheap and burning 🙂

    Kind regards

  154. Torgeir Fredriksen

    Mads, if you ever were to make a LCLR kit that run stable, please add a comment here. I would highly appreciate being notified about such an initiative 😀

  155. Hi Torgeir

    You can be absolutely sure that all projects that I succeed in making work will end up documented on this website, it is another story with all the failures, not enough time to write about that 🙂

    Kind regards

  156. Hi Mads,

    Thank you for the reply. I actually doubt my own power supply as well, so I replaced such expensive power supply in my store (I couldn’t sell that one) with an old PC’s power supply. It means that I have to run the circuit in lower voltage (12V instead of 24V), but it can support more current (up to 21A). This time, the circuit works! It isn’t sufficient to heat things up, however, but I could see a beautiful sine in the oscillator.

    Now I am working to make a stronger circuit with my new transformer. While I am at it, I want to improve a lot of things as well. In the proof-of-concept circuit, the capacitors tended to heat up pretty fast, so they ended up bursting or degrading after around one minute working. I was using some cheap CBB capacitors (dunno which manufacturers) with normal air cooling, so it was expected, but is there any tip to keep them safe aside from good cooling (I will equip water cooling with an old pump)? I have some secondhand WIMA MKP10 capacitors, are they decent enough? I saw some circuits on Google using really big one, so should I invest in them? Would some old, degraded fan capacitors work?

    As for the final circuit for my usage, I am aiming for kW milestone. I haven’t had my transformer yet, but I will be using a 3kW one with one output is 60V (I will also use this transformer for welding as well). Is the maximum current 50 amp achievable for the zvs? Would such current fry the IRFP250 up? Should I aim for more voltage to ease the current? I quite hesitate to raise the voltage over 60V as I will be using a MOSFET with 200V+ rating, which is, as you said in one of the comment, unsuitable for the zvs heater. Also there is this site saying that the zvs driver will suicide with 70V+ supply:

    Once again, thank you for the reply.



  157. Hi Ahirube (afk)

    Capacitors for induction heating is usually water cooled and heavily heat sinked for industrial applications. For reliability you need good quality and the same goes for a kW range induction heater, if you want to actually use it to work for blacksmithing, you need better capacitors with much higher RMS current rating.

    WIMA MKP10 capacitors are in the good range of small film capacitors, but are not suitable in the long run, you can not sufficiently cool them enough.

    IRFP250N is only rated for 30 A continues drain current, you will have to look at your total conduction time at your resonant frequency and look up what current can be conducted in the datasheet graph for safe area of operation.

    Adam knows what he is talking about and I have it described in the top of this article first paragraph in considerations.

    Kind regards

  158. Halo Sir,
    I have some clarifications on frequency of LC circuit. If the frequency is lowered to around 75 or 80 khz by changing the inductance and capacitance values, will it take more time to heat the metal to red hot?
    Does frequency plays a major role in the time, taken to heat the metal to red hot?

  159. Hi Prakash

    Metals heat up most efficiently at different frequencies to different metals, if you look in the file archive of my site, find it in the menu, there is some documents regarding induction heating.

    Kind regards

  160. Hi Mads,
    Thanks for your kind reply Mads.
    I have an other question… Why do we have four electrolytic capacitors at the rectifier side i.e why do we need 4*80vdc/10000v caps. These caps are for filtering and am clear with that but what is the criteria for using 4 caps instead of having 1 cap and how to assemble these four caps in the rectifier side?

    Thanks Regards.

  161. Hello Mads,

    Thanks for descibing this project. What kind of rating should the smoothing capacitor for the rectifier have, just one 40mf capacitor or multiple in series? How much current will this/ these capacitor(s) see?

  162. Hi Joost

    I had 4x 10000 uF in parallel to have enough capacitance to lower ripple current and to be able to conduct the current used by the circuit. It was around 20 A drawn from the DC bus here.

    Kind regards

  163. I have been trying to figure out how to use induction heating coils on large objects. My intended application is a regular 10 liter stainless pot and I would like a very large coil surrounding the outer diameter of the pot. Will the large diameter of the setup require too low frequency for this project to be feasible or are there other problems associated with this kind of setup?

    My application would only see temperatures at or just above 100°C and I plan to use water cooling of coil, caps and switching components.

  164. Hi Mikael

    I would suggest that you make it like a regular induction stove, with a flat pan cake coil at the bottom, I do not think there is too much trouble with the heat distribution when you are just around 100°C. It is not as much the lower frequency that is a problem, it is more that the area over which you heat with some 500-1000 Watt gets too large. There might also be problems with finding a suitable capacitor that are still within a few uF without dropping frequency even further.

    Kind regards

  165. Hi Mads..
    Actually what’s the purpose of having two inductors L1 and L2 in the circuit?
    I have only 100uh 6A coil of 5 pieces with me and will it satisfy my circuit?

  166. Hi Prakash

    Two inductors are needed to use a single wound coil and not a center-tapped as originally used in the Royer oscillator.

    Kind regards

  167. HI Mads
    Thanks for the kind reply…does the effective inductance is around 65uh while considering L1 and L2?
    And I have 100uh coil of 6A each…can I parallel these five pieces to get 30A rating but by doing that I loose effective inductance …what is the solution to this…


  168. Hi Prakash

    Best would be to find some suitable inductors, but you could also try to use two inductors in series and parallel two of those strings. That would give you 100 uH at 12 A, good enough to try it out, but will properly only work for short runs before heating too fast.

    Kind regards

  169. Hi Mads,

    I have a beefy heatsink that I want to use for my MOSFETS. Can I mount BOTH mosfets to the same heatsink in this circuit? Looking at your photos, it doesn’t look like you’ve got them electrically insulated from each other.


  170. Hi Joost

    You need to isolate the heat sinks from each other! If you look closely at my pictures you can see that it is two heat sinks glued together to a piece of acrylic plastic.

    You can also use isolating pads and washers if you want to mount them on the same heat sink.

    Kind regards

  171. Thanks. I just now read that exact freaking thing in the description under the picture…
    Too bad, I had the perfect heatsink to mount these two to (don’t have the insulating stuff)

    Guess I’ll cut up a smaller heatsink

  172. Hi Joost

    You did not miss it the first time, I updated the article right after I wrote the reply to you 🙂

    All the questions asked here are used in perfecting the articles with more details.

    Kind regards

  173. So I build the device and tested it. I got a screwdriver luke-warm before a MOSFET blew. The entire “circuit” is a rats nest, and a complete mess (I’ll fix it when I get it running reliably) so I thought the heatsink might’ve shorted with the capacitor bank. I then replaced the MOSFET with a spare and tried again. No blue smoke, but the circuit didn’t work at all this time. I’ll check the diodes and such, but is there something from this description that makes you suggest a likely fault? The mosfet blew at the source.
    I use IRFP250N’s, imput is about 50 volts from a rewired MOT, Full bridge rectified and 40mF smoothed.

  174. Hi Joost

    Resistors and zeners at the gates of the MOSFETs can also get damaged from a blown MOSFET, be sure to check those too.

    What is the voltage and current at full load? It could be that the Royer pulls the voltage down under 10 VDC and it latches up or the voltage is too high. 3.14 x 50 VDC + switching spikes could be above the voltage rating of the IRFP250N.

    Kind regards

  175. Hi Mads,

    I can’t measure the full current load, my multimeter can’t take the heat. I could check voltage but I’d have to wire it back up again. The IRFP250N is rated at 200 volts. I could rewire the MOT for a ~40V input.

    For reference, my parts list:

    25mm cilindrical ferrite core (x2)
    10000uF 63V Radial Electrolytic Capacitor 105C (x4)
    KBPC2502 25A Bridge Rectifier 600V
    Diode Ultrafast UF5408 Gleichrichter 3A 1000V < 75ns
    WIMA Polypropylen Folien-Kondensator MKP4 10% 250V 0,1uF (x40 in paralel to get 4 mF)
    in5349b zener, 3.3-200V, 5W
    2W 470 and 2W 10k resistors. I increased the resistance for the 470's by an aditional ~260 ohms by wiring up 8 .25W resistors and one spare 2W 470 resistor in paralel, and then putting that bunch in series with the original 470. I did something similar for the other 470, but that total resistance is maybe 20 Ohms higher. Shouldn't be a problem right? If I bring the voltage down to 40, I can just use the one 470W.

  176. Just found out the 2nd MOSFET is cracked too. New parts ordered. I will rewire the MOT to about 40 volts. I also ordered 620 ohm 2Watt resistors.

  177. Hi Joost

    Your parts list look good, you chose good quality components. When ordering resistors for gates, they are usually cheap and thus it should be no problem to order 10 of them and find 2 of them that are most identical.

    MOSFETs can also die silently, without any cracks, measuring them with a multimeter can most often reveal if it is good or bad.

    Kind regards

  178. Hi Mads and thanks for your reply

    I actually tried a coil wrapped around a stainless pot. I bought a 1000W ZVS from ebay and wound a 7 turn coil around my pot (about 5 meters of copper tubing). I still haven’t managed to measure the current draw but it is a LOT! I also noticed that you need a hefty power source the get the oscillation going so I took a large UPS battery and it seem to do the trick. The pot gets warm in just a few seconds and on my oscilloscope I can see an almost perfect sine wave over the working coil. The original ebay-coil together with the capacitor bank resonates at 91kHz but with my large coil I get round 60kHz instead.

  179. Hi Mikael

    That sounds great, just be sure to check temperature rise of the components on the board. The China 1 kW ZVS boards are in many cases of worse quality components than the DIY.

    Please attach pictures of your setup, I would like to see it 🙂

    Kind regards

  180. Just scored a clamp meter. The current draw is 30A @ 12V so roughly 360W. I honestly thought it was more but 360W aint bad. The plan is to use a 24V PSU and that will probably get me close to 750W with the current configuration but I think I need double that.

  181. Hi Mads,
    I want change something of them: Mostfet IRFP250N (200/30A) to IRFP460PBF (300V/20A), Source 50VAC ~ 70VDC -> 280VDC 2,2mH, C: 1,98UF, => F: 76 Khz, L1 and L2 :30T with wire dia 1mm. Is it can work 1500W continuse? If don’t, How can I do that? (1500w)

  182. Hi Mads,
    How can i change that one to direct 220VAC (not transformer), 1500W or more?

  183. When I removed one turn from the working coil I got 28A and 77kHz instead. I plan to order some litz wire on ebay soon and give that a try. I think it will give me a bit more power and also probably easier to bend than my copper tubing. Should I aim for high or low frequency for my desired application? How big part do the two chokes play in the efficiency and working frequency of the ZVS induction heater? When studying schematics of a similar design it calls for 47-200uH but I suspect choosing the right value makes a lot of difference, am I right?

  184. So this time the circuit lasted about 30 seconds before a MOSFET blew out. It did heat up a screwdriver to smoking hot before doing so. Rewired the transformer again to 38 volts (down from 42 and a bit). My work coil has 6 windings, I’ll remove one and try again. The (solid copper) wires going to the coil get pretty warm, and so does the transformer.

  185. Hi Mads,

    Ok this isn’t working for me. I’ve changed the voltage 3 times now: I’m down to about 32 volts and I still can’t get the device to run for more than about 90 seconds before a MOSFET blows. The MOT is actively cooled with a small fan, the rectifier is passively cooled with a big heat sink, and the capacitor bank and the MOSFETs are actively cooled by a big 12v fan, and the MOSFETs themselfs are each mounted on a heat sink. Even after 90 seconds, I can’t get a screwdriver to glow red hot. It gets hot, but not that hot. I’m starting to think that those MOSFETs just aren’t up to the task and that I need bigger ones.

    What else could be wrong here? I’m using 25mm ferrite cores with 22 windings of 1mm copper wire. Instead of 470Ohm resistors, I’m using 620Ohm, I have 40 100nF WIMA MKP capacitors in paralel for 4mF. I use the same diamater copper coil as you with 5 turns. Considering the circuit actually works for a while, I asume everything is hooked up properly. What kind of power supply are you using?

    Any suggestions as to what might be wring would be very appreciated.

  186. Hi Mads,

    Thank you for making this video. I just wanted to give you an idea where you can get good capacitors for your induction heater experiments – take an induction stove from the scrapyard 🙂 There are units with 1-2kW output power, so you can easily heat up a much bigger iron piece than nails and screws. Just don’t forget to run water through the coil, which in this case MUST be made of tube. Preferably a thick-wall one.

  187. Hi nguyen nhuan

    This circuit scales up best to those power levels by having multiply bridges in parallel, with each their driving circuit, see these links for how Roger did this: and

    This circuit does not work well with high voltage input, consider 320 VDC in results in over 1000 VDC on the switches, they become too slow and has too small current rating. You will have to move on to more sophisticated control methods to get rid of the transformer.

    Kind regards

  188. Hi Mikael

    Lower resonant frequency is better at higher currents and voltages, your switching losses will be less. But you should research in the optimum frequency for heating the material you have in the coil and let that be a design criteria.

    Inductors in a DC circuit stores energy and it resists the change of current. It supplies energy after cut-off and that decays fast. Large inductance does however also result in a larger voltage drop across the inductor.

    Kind regards

  189. Hi Joost

    Does your voltage drop below 10 VDC? This is a common reason for the Royer circuit to blow up. You load the transformer with so much current that the voltage drops, as it can not supply more power before going into saturation.

    Kind regards

  190. Hi Ivaylo

    Thanks for sharing this tip with all the commenters. I think I already got parts from 6 stoves in boxes, that I found at a scrap yard.

    Generally any MKP capacitor of good enough voltage/current rating can be used.

    The stove electronics are however prone to failure, they are just not cooled well enough in their original design. It has gotten better over the years, but I have seen many that are blown or burned.

    Kind regards

  191. Hi Mads and thank you so much for answering all our questions.

    So when it comes to the DC chokes, bigger is better? Would something like this monster work?

    I’m guessing the voltage drop only depends on the DC resistance of the windings so ridiculously over sized wire gauge should take care of that?

    I have made a couple of different, much smaller coils from the copper tubing, and one 4 turn coil resonates at about 160kHz with a small piece of metal inside the gap. I got the piece glowing in about 30 seconds. Also Litz wire is on it’s way. Should be fun to see how that changes the power.

    Again thank you. I will publish my findings and measurements here as the work progresses.

    Best regards

  192. Hi mads,

    I have not yet tested this. I use a 1kW MOT so you’d think it could handle 750W. If this is the problem, how can I prevent this from happening, wire up a second transformer in paralel to split the load for example?

    What kind of power supply are you using?


  193. I have 2 MOSFET’s left with questionable integrity. I’ll check for voltage drop. Thanks for your help!

  194. Hi Mikael

    Bigger is not better!

    Certainly not at all when this one has a iron core, it will saturate long before reaching the switching frequencies in a Royer oscillator like this. You need a ferrite core that works well within your resonant frequency.

    The voltage drop in a inductor is a result of the opposing magnetic field to the current flowing through. DC resistance is negligible. The larger inductance the larger voltage drop and you might end up with a severe voltage drop in this low voltage application.

    Kind regards

  195. Hi Joost

    Microwave transformers are a black magiv box in electrical engineering. It is designed to operate right at the brink of saturation and has metal shunts in it to limit the current. They are simply undersized for their VA rating.

    I used a large 1000 VA transformer, its properly about 4 times heavier than a standard MOT.

    A second MOT in parallel would be a good solution. But try to measure the DC voltage under load.

    Kind regards

  196. The MOSFETs I have left are broken. The transformer I use is rated for 1kW and weighs 5 kg. I can always get rid of those shunts

  197. Alright new MOSFET’s, new ways to blow them up. I measured the voltage across the rectifying capacitor bank to be 33.9 volts, unloaded, so no circuit attached.
    I then hooked up the circuit, stuck a screwdriver in the coil and measured again. Voltage drops to about 25. I’m still measuring the retifyer. So that seems to be in order.
    What about the inductors? I used 1mm diameter copper wire, 22 windings. Could that be a bottleneck perhaps? How about the heatsinks? Mine are smaller than yours.

  198. I shorted out the transformer again. I removed the shunts and I’ll try thicker wire.
    Next I’ll measure voltage across the gates

  199. I actually removed the shunts and rewired the transformer with solid, 1.6mm copper wire. Now however, the transformer primairy heats up even if there’s no load! just measuring voltage for like 20 seconds makes the primairy luke-warm

  200. Hi Mads,
    One of the reputed industries asked me for an induction heater for forging machine which must be run for about 7 hours a day(according to the orders it varies to 3hours or 6 hours a day). So will this design suits for the indutrial purpose.
    Thanks regards.

  201. Hi Prakash

    No, this is a highly unstable circuit and it only suitable for experimental use. It will be cheaper to build a proper LCLR induction heater than trying to make this circuit bullet proof.

    Kind regards

  202. I measured inductor and capacitor bank values: capacitor bank is at 2.4 microfahrad, inductors are 2.5 MILIhenry, WAY too much. I now have 22 turns, it should have something like 5.

  203. Hi Mads,
    The capacitors at the rectifier side always pulls more current. The capacitors are 10000uf/63v of 4 pieces each connected in parallel. Actually I have made the power supply by AC voltage control using triac and that controlled AC voltage from 110v to 45v is been rectified by rectifier module and finally the capacitors. Please help me out with the capacitors which pulls more current and damages triac. Without capacitors I get the 43v with lamp load(just for testing I used incandescent lamp), but with the ripples.
    Thanks Regards.

  204. Hi Prakash

    A empty capacitor looks like a dead short circuit to a power supply, you could use a soft start circuit where you initially charge the capacitors through a resistor or lamp, and after a few seconds you switch out the resistor with a relay. It could also be that the induction heater just draws so much power that your triac regulation is not powerful enough.

    Kind regards

  205. Hi again Mads

    Would this capacitor be a good choice for the Royer oscillator or is the circuit just not suitable for high power?

    Also I read your info about DC bus caps and the picture you made of how to properly parallel several caps – but is this also applicable in the tank circuit if one were to use several smaller caps instead of something similar to the Celem?

    How is it that even professional high power induction heaters don’t use Litz wire but instead copper tubing? Wouldn’t Litz be more efficient?

  206. Hi Mikael

    Celem makes the best induction heater capacitors you can get for money, so they are expensive. This one you linked is a 800 kVAr rated capacitor! That is about 1000 times higher power than you should expect to get through this circuit 🙂

    Yes, you should distribute capacitors in parallel the same way to ensure correct current sharing!

    Tubing is a thin wall, so with skin effect taken into account you actually use almost all the copper to conduct current in. Tubing is super easy to water cool, litz wire is not! The latter being the most important parameter.

    Kind regards

  207. Thank you for your answer Mads.

    A little side note… about the 470 Ohm power resistor that handles the gate charge… on the ebay models there is a wire wound power resistor that probably has a quite a bit of inductance. Wouldn’t a metal film resistor with adequate power rating work better due to the fairly high frequency?

  208. Hi Mikael

    You are absolutely right about wire wound resistors are ill-seen as gate resistors, but given its high resistance value, it does not pose a great problem here. I would also prefer carbon or metal film resistors.

    Kind regards

  209. Ok, bare with a few seconds. I may be WAY of base here bout would it be possible to make a water cooled capacitor with SMD ceramic caps rated at 200V?

    I plan to make a simple reflow soldering oven in the near future. Probably controlled by an arduino. If I were to place let’s say 200 small ceramic caps between two of these

    and perhaps even reflow solder the mosfets to the other side of each water block… wouldn’t that make a cheap induction circuit that only needs a few external components?

  210. Hi Mikael

    That is a exciting idea, I would however go a capacitor rating around 4-5 times your supply voltage.

    I hope you can pull this off and show of the result 🙂

    Kind regards

  211. I plan to use 24VDC as supply voltage and smd caps with voltage ratings above 250V are pretty much impossible to find. 200V should give me a margin of almost 100V but I guess I could just as well use the 250V caps since they are pretty much the same price. Couldn’t hurt.

  212. Hi again Mads

    I received a thin Litz wire the other day that I ordered from ebay a few weeks back. It was 180 strands of 38 AWG wire and I made a small coil (5 turns) around a small alu tube I had lying around. A borrowed 600W SMPS started to feedback at 40 Amps so I took a large lead acid battery. Didn’t check my clamp meter but it was probably a lot of power and I managed to fry on of the mosfets so I am almost back to square one. I think I have concluded that the skin effect only occurs on the outer surface of any conductor so if I want to transfer the generated heat to the liquid inside the pot/working piece I will probably require a low frequency induction heater. A lower frequency also penetrates deeper so I assume that a larger area affected also means more energy transferred? I need to find some 200A fast acting fuses to save my mosfets 🙂

  213. Hi Mikael

    I got no experience in running it at audible frequencies, for the obvious reasons that it is in the audible range and would be annoying as hell to be near.

    Kind regards

  214. Also for the power supply circuit, using the autotransformer with 2.5kva power rating with input 230v amd output 0 to 230 volts and maximum primary current is 9 amps secondary current is 50 amps so the net secondary current is load current minus line current is equal to 39 or 40 this okay with power supply circuit excluding recrifier circuit

  215. Hi S. Prakash

    As you can see in the video, it is just about under 1 minute to heat the bolt to red hot.

    A 2500 VA rated autotransformer should be suffecient, just remember to check the rating curves, not all variacs can handle a high current at low voltages!

    Kind regards

  216. Hi Mads,
    Thanks for your kind reply. Does the resonant frequency is related with the rate of speed of heating the metal to red hot condition? By improving the resonant frequency we could increase the speed of heating time from 60 second to 45 seconds or less than that?
    Thanks regards.

  217. Hi Prakash

    Check out my file archive, link out in the menu to the left, browse to induction heating folder and there are some old scanned books about induction heating, there should be some mentions of optimal frequencies to different types of materials. What really matters in heating speed is the energy transfer, a higher power heater will be much faster than optimizing frequency.

    Kind regards

  218. Hello sir,

    I don’t know much about electronics, but I would like to build this induction heater and I didn’t unerstand some things from the coments. So therefore I would like to ask you if:

    1. Could this induction heater work with no inductors at all? From this youtube video in the following link : I understood that there is no need for an inductor, because he has a permanent power supply (maybe from a transformer).

    2.Can I use electrolytic capacitors (combined maybe) in C1? Maybe just electrolytic capacitors? of course respecting the polarity. And for theese non polarity capacitors can I combine just any capacitors to get around 2 to 4 uF? Or do they need to resist to a high voltage that is written on them?

  219. Hi Marian

    In the video, the first component you see to the left, red ring core with copper wire wrapped around it, that is an inductor.

    You can not use electrolytic capacitors in a resonant tank circuit as this is. They would be destroyed in seconds. You need high power film capacitors like described under considerations in the start of this article.

    Kind regards

  220. Yes Mads ,but in that video in the link I sent you, that red ring core with copper wire wrapped around it the inductor -> it said that there is no need for it , because it has current from a microwave transformer (maybe the inductor would be neccesary if it had a bettery power supply). So therefore if I have current from a transformer connected to our circuit ( the one on this website) there is no need for inductors right?
    It looks hard to calculate the micro Henrys, and even the cheepest inductance meter is kind of expensive.

  221. Hello again Mads

    I am sorry to bombard you with questions but I’m about to order solder paste, zener diodes and other components I need and I was wondering what would be the best type of diode for “400V fast diode” ? I can choose basically anything from Farnell and I don’t mind paying 10-20 USD each if it does the job better than a cheaper diode. Should I choose a ultra fast recovery “regular” diode (I have NTE519 in my current shopping basket) or would some kind of TO-220 schottky work better?

  222. Hi Marian

    The inductance of the inductor is not critical, so you do not need exactly calculated, but merely as described a number of turns is roughly in that ball park.

    I have not tried to run this circuit on AC, so I do not know if the inductor can be omitted in that case.

    Kind regards

  223. Hi again Mads

    Where do you actually connect the two ends of the work coil? In the video from this link : I saw the ends of the work coil connected just to the two heatsinks at the end of the video. You know in this video that heater had exactly the circuit diagram that is on this website. Now in this case if the work coil is thus connected the mosfets would not have to be electrically isolated from the heatsinks right? And how do you stick the mosfets to the heatsinks in this case?

    P.S. I do know how to read cyrillic (if this could help), but I don’t understand what that guy says right there. I don’t even know what language he speaks.

  224. Hi Marian

    Good observation you made there, the tab/heat spreader/metal back of a MOSFET is internally connected to the drain pin and thus the two MOSFETs needs to be mounted on two separate heat sinks in order to use it as you saw here.

    If they are mounted on the same heat sink, insulating pads, plastic screw protector etc is needed when mounting it to avoid short circuiting.

    Kind regards

  225. Hi again Mads,

    I read that the thikness of this work coil (copper pipe) is 8mm . But what is the diameter of this work coil that you put things in? Around 1 inch?
    And does it matter if this diameter is + or – 0.3 inches?
    I would like to make this work coil have a as thin diameter as possible because I don’t know if my copper pipe is long enough to make it have 5 turns.

  226. Hi Marian

    The work coil geometry and size is always dependent on the work piece you need heated. But you should use my helical coil calculator to ensure that the frequency does not get too low or high with the design you intend to use.

    +/- 0.3 inches should not make any significant change in operation.

    Kind regards

  227. Hi Mads,
    Is ter any problem if I use 30mm dia rod that has to be heated in the 60mm dia work coil?
    At first I used 7mm rod for heating and that worked fine but when I kept 30mm rod inside a 60mm dia work coil the mosfet blowed. What would be the reason.

  228. Hi Prakash

    A 30 mm rod will absorb much more energy and therefore represents a much bigger load to the circuit, remember that there is no current limitations in this circuit so it will draw as much as its power supply allow it to or just pull down the voltage while doing so.

    Kind regards

  229. Hi Mads,
    Ya I got it and to make this work well , shall I use the low inductance value designed as 0.299uh so that my resonant frequency would get to its higher and reasonable value. This could avoid my circuit getting blowed?

  230. And also by using the stainless steel coil as work coil will improve the heating effect though the resistance of stainless steel is higher than copper and brass?

  231. Hi Prakash

    Nothing can keep this simple circuit from blowing up, it is by nature unstable as there are no safe guards to avoid dangerous situations for the MOSFETs or capacitors.

    A higher frequency will result in increased switching losses and would require better cooling of MOSFETs, capacitors and work coil.

    Stainless steel should never be used as a conductor, its high resistance would limit the current flowing in the work by so much that you would not be able to heat anything. Remember that the heating effect is a transfer of current from the work coil to the work piece in a say 5:1 ratio.

    Kind regards

  232. Thanks Mads….can you suggest me any of the circuit that suits to heat 30mm rod for industrial purpose..I saw a few infuction heater circuits like H bridge circuit and few but little confused and so please suggest some of the kind that you know.

  233. Its okay Mads….by using high frequency tank circuit is it possible to heat that 30mm beast without any problem and dont consider about mosfets in this case. Will high frequrncy circuit will provide the necessary energy to heat the 30mm rod?

  234. Hi Prakash

    Frequency does not play as important role in induction heating as actual power transfer does. Do not expect this circuit to deliver more than around 600 Watt, maximum 1000 Watt.

    If you want a industrial sized induction heater, you have to look into building a LCLR induction heater instead.

    Kind regards

  235. Hi Mads,

    I might bombard you with some questions if you don’t mind. I have build the induction heater (as you can see in the images from the links bellow) but it doesn’t seem to be working. I touched the work coil but it doesnt even get warm. I’ve put some metals in the work coil but still didn’t get hot . However the heatsinks attached to the mosfets got pretty hot. Now if you look at the images in the links below can you tell me what did I do wrong? First of all did I connect the work coil proprerly?

    -I used 2 x 2.2uF (film or polymer capacitors I don’t know ) in parallel in C1. This means 4.4uF in C1.
    -The inductors are in series and they don’t touch any other component in the curcuit.
    And I don’t think that there would be a problem with any other components on the circuit board. The smaller circuit board behind is the AC to DC converter which suppleis power from a transformer. I supplied this circuit with around 15.6 V DC.

    Thanks, Marian

  236. Hi Marian

    The Royer induction heater will eat almost all the power it can before either pulling down the power supply or conducting so much power that it burns up, there is no inherent current limiting except what is limited by the current ring up in the primary circuit and what energy can be transferred to the work piece.

    With a 32 VDC supply and pulling over 20 Ampere, you have the estimated 650 Watt power that I demonstrated.

    Your very small power supply is not even enough to drive the gates of the MOSFETs, if you try to measure the DC voltage you will properly see that it is much lower than 15.6 VDC because the load pulls the voltage down to a point where the MOSFETs are only operated in linear mode and are just burning power off in themselves.

    Another thing, you can not use test leads for this setup, you need heavy gauge wire that can handle the large supply current and the primary circuit that consists of C1 and work coil needs to be able to conduct hundreds of Ampere at high frequency.

    Kind regards

  237. Ok Mads , you said that if I try to measure the DC voltage I will properly see that it’s much lower than 15,6 V DC. Now where do I have to put the pins of the multimeter to see that this DC voltage is much lower? On what parts of the circuit board? Or in which points the voltage is pulled down? Before connecting the the power supply to the circuit board, I measured the volts coming from that AC to DC converter and it showed like 15,5 V DC.

    Now the capacitors combined in C1 are good?, they just have to to conduct hundreds of Ampere at high frequency right?
    And how thick would this heavy gauge wires have to be? Could I not attach the heavy gauge wire to theese aligator clips which have screws? I mean to tighten the heavy gauge wires with the screws. You know like this I could easily dettach and reattach the work coil to the circuit board and it would be more compact. But if I can’t it’s not such a big problem.

  238. Hi Marian

    You seem to lack a basic understanding of electrical resistance, current and so on.

    The supply voltage would show a drop at the source, so where you measure it now, would be the same place when its connected to the circuit.

    A single capacitor like those can maybe conduct a few Ampere and those two in parallel, maybe 10 Ampere and they would start to heat up. Hundreds of Ampere RMS current is no joke, you need heavy wiring, current values you could find in my wire size table in the bottom of the menu.

    Alligator clips have absolutely no place in a circuit like this, they are for low voltage dc playing or logic level test setups. They would melt almost instantly if used in the resonant primary circuit of a induction heater.

    Kind regards

  239. Yes Mads like I said I don’t know much about electronics.

    I thought to put my capacitors in C1 like this in the following picture Each capacitor having 2.2uF. Maybe I’d use a separate circuit board for C1.

    If I arrange them like this would all these capacitors resist theese hundreds of Amperes?

    I’m not going to use alligator clips. But how will I connect the work coil to the circuit board then? As you could see in the pictures from the links I sent you my work coil (copper pipe ) is barely long enough to reach 5 turns. Could I use thick gauge copper wire connected to something like this in the following picture , and then screw the copper ends to my work coil?

    And how did you connect the capacitors to the work coil with solder? This is what I saw from the pictures on this website. And if you connected them with solder wouldn’t the solder melt ?
    And where is your wire size table at the bottom fo the menu? I didn’t find it on this website. Can you give me the link?

  240. Hi Marian

    You have to look up the current ratings of these capacitors in their datasheet, to see how much current they can conduct at your resonant frequency. You could alternatively look at my MMC (multi mini capacitor) calculator, I have those red capacitors in the list with pre-entered data.

    The capacitor arrangement look fine, just be sure to install the lead in each their end of the parallel connections, so the upper should be connected out in the right side and the lower out in the left side, this is to ensure a better current sharing. If you look at the length of the path the current has to travel, the path would be shortest through the middle of your illustration and the capacitors in the middle would properly conduct 80% of the current and get damaged.

    Sure, you can use home-made cable lugs from copper pipe to mount a large wire to the work coil. Solder will first melt at some hundred degrees Celsius, so as long as your wire and capacitor is rated for the current that it conducts, it will not heat up enough to melt the solder.

    Wire size table:

    Kind regards

  241. Hi Marian

    They are properly identical to the Panasonic capacitors: which you can also choose in the calculator:

    The left solution is correct, I described this in more detail in my DRSSTC guide, see this chapter on capacitors (scroll down to “Current sharing between capacitors in parallel”):

    Kind regards

  242. I’m going to power this circtuit with 32 volts or more as you said, but I can’t get that much voltage from only one transformer? I could put theese many batteries in series but even then wouldn’t this heater consume much energy and the batteries run low very fast? I might need a permanent power supply. What can I do to increase the voltage? Just as I combine the batteries in series and add up their voltage, can I not increase the DC voltage with two transformers and add their voltage like in the image bellow?

    You know if I combine the transformed AC voltage and then connect it at the DC converter that electrolytic capacitor may explode or no longer be good because it says it resists a maximum of 16 volts.

    I’ve looked on the internet but I just couldn’t find the answer if the voltage from the transformers could be thus combined. So if this idea works or not I’d suggest that you would publish this image in the internet so that it could be easily found in goolge search.

  243. Hi Marian

    You need a large transformer, in the region of 1000 VA rating, small power supplies like the one you already showed will not work.

    You can put DC supplies in series to get a higher voltage, but you can never get more current than what the smallest power supply can deliver.

    Kind regards

  244. Then what power supply can I use? I’ve got 8 batteries of 4.5 volts. They may run low pretty fast, but would they give enough power?

    And you said that almost any regular diode with a ultra fast description will work fine. What is a diode with ultra fast discription? Any diode with UF written on it means it is ultra fast and it’s good?
    I sure hope my mosfets are still ok beacuse the heatsinks got so hot that I couldn’t touch them. How do I know that my mosfets are still good? Can I test them without taking them out of circuit with this method in the link below?

    Or if the heatsinks get hot (or warm) again when I turn on the heater that means the mosfets work just fine?

  245. Hi Marian

    Some LIPO or LI-ION battery backs could properly supply enough current for this circuit, but I would not rely on any batteries, except maybe two or three car batteries in series for 24 or 36 VDC, but be aware that they can supply hundreds of Ampere and if something goes wrong, your circuit will exploded.

    Ultra fast diodes are some kind of loose “industrial standard” for naming diodes that have a certain low reverse recovery time, check out some UF diode datasheets to see what the times are to compare to others.

    I do not recommend to test a component when it is in a circuit, you got no idea what you are testing then.

    Heat sinks get warm in this Royer oscillator, but that is absolutely no sign of it working, if you have oscillations in the primary circuit and are able to heat a work piece, then it is working.

    Kind regards

  246. Hi again,

    I asked you before about the diameter of the work coil , but does it matter how thick is this work coil (copper tube)? Here on this website it says its 8 mm thick. Could I use 15mm thick? Maybe the water could better flow through that one.

    And the peace or meal that you put in the work coil does it get levitated like in the video from the link below no matter what metal it is?

  247. Hi Marian

    You can use a variety of different tubing width’s, but it will affect your resonant frequency, so you have to recalculate that for each work coil and see if it is viable to use for your induction heater inverter.

    It takes a powerful induction heater to levitate metal and you need a work coil that is counter wound on the middle in order to trap the work piece between to opposite magnetic fields.

    Kind regards

  248. And where do you find this formula for calculating the tubing widths?
    And what temperatures could this induction heater reach? I thought to melt thungsten filaments from light bulbs which melt at a very high temperature, will it melt them?

  249. Hi Marian

    You just measure the width of your tubing with a ruler or caliper.

    Induction heaters if low power, like this, will only be able to heat metals up to their curie point, from here on electromagnetic heating efficiency drops a lot and it now takes much more energy to melt metals. This heater will not be able to melt tungsten.

    Kind regards

  250. Hi Mads,

    First of all, thank you for having the time to answer all of these questions 🙂

    Secondly, I am trying to design an induction heater to operate around 2 MegaHertz. I understand that the frequency will change the LC components drastically, but would your design be able to withstand this increase in frequency? Hoping to supply around 100A of current through coil for heating small particles.

    Any help or suggested design techniques are greatly appreciated


  251. Hi Tyler

    You need some really fast MOSFETs to be able to switch at 2 MHz without having excessive swtiching losses. Gate drive of the circuit might even be too weak. I think you will have trouble finding a fast enough MOSFET to work at that frequency, with 100A in the primary circuit.

    I think that high frequency induction heaters usually uses vacuum tubes instead of MOSFETs/IGBTs. Maybe you should try to look into that.

    Kind regards

  252. About this thikness of the work coil I wanted to say that I might replace this 8 mm thick work coil with a 15 mm thick work coil. What circuit modifications or changes should I make after this replacement?

  253. Hi Marian

    I do not think you have to change anything, but be sure to check what frequency the LC circuit will resonate at, if it is either half or double or your original, then I would say you have changed it too much and might have to add more capacitance to the tank capacitor.

    Kind regards

  254. Sorry for bombarding you with too many questions Mads, but building this heater is a lot harder than I thought.
    Now I have some microwave transformers (in the links), can I use them to power up this induction heater? Of course converting AC to DC. Would only one transformer be enough? If not I may put two transformers in series maybe two would be enough.
    And would theese magnetic shunts eat a lot of the transformers power?

    The primary coil I think it has aroud 220 turns powered at 220V AC, and at the secondary I did wind 5 turns high and 3 layers deep. Can I measure the voltage at my secondary coil with my cheap multimeter?(in the link also) The secondary coil could have too many amps, could it damage my multimeter?, or when I measure the AC volts, the many amps won’t have nothing to do with my multimeter. My cheap multimeter won’t measure the AC amps, only DC and only up to 10 amps.

    And when I convert the AC to DC current from this transformer/transformers what diodes and capacitor/capacitors should I use?

  255. Hi Marian

    I do not mind helping, that is a part of releasing all this information for public use and education 🙂

    Yes, you can use rewinded microwave oven transformers, but make sure that you have at-least 24 VDC, I recommend that to avoid too much voltage sag doing heavy load. You can configure the transformers secondary outputs in both parallel and series connection for the voltage/current needed.

    You should remove the shunts, to get rid of the current limit and to give you more space to wind in.

    You can measure the output voltage with your multimeter, no problem there. But you should not try to measure current with it in series, that will highest likely blow the 10A fuse.

    You should use one or two 25A bridge rectifiers ( like this ), best would be two in parallel, mount them on heat sinks. For filtering you should have at-least 10000 uF, the more the better 🙂

    Kind regards

  256. Hi Mads, thank you for the response again 🙂

    My curiosity for high frequency heating was piqued from this article/topic where someone had appeared to construct a 5Mhz heater. I understand the speed of the FET has the greatest impact on the circuit due to switching speed and losses, hence the need for a heat sink most likely, which does mean that the efficiency and power usage will still be high. Do you think the ZVS would minimize these losses?

    Link to 5Mhz heater:

    Warm Regards,

  257. Hi Tyler

    Yes I remember that thread on 4hv now 🙂

    I think a part of his success lies in it being a single transistor solution. It can not short another transistor out and explode.

    I think the problem will be adequate gate drive and cross conduction in the Royer circuit, but hey it is so cheap and easy to make that you should try it out and see what happens 🙂

    Kind regards

  258. About powering up the induction heater with theese transformers; the transformers I will put them in parallel because I don’t have a higher voltage souce to put them in series, but can I put in series the secondary coils as you can see in the link below?

    Now how do I connect the secondary coils in series? The phase of one coil with the neutral of the other?
    I searched on the internet, but it’s very hard to find the answer to this question so I think you are the one who could give me the fastest answer.

  259. Hey Mads,

    I tried it out, and I think I got the tank circuit to oscillate with a 1uF capacitor bank, and a work coil with ~1uH inductance at 160KHz.

    My question is now, as I increase to the desired frequency either the work coil or capacitor bank must also decrease, the work coil sort of has to stay the same. But I was wondering about the two choke coils, currently i have 2 coils both at 150uH, would the value of these have to decrease for higher frequencies?

    PS I double checked the rise/fall times expected of the FETs I am using, and theoretically they should be able to handle the high switching speeds 🙂


  260. Hi Tyler

    Good that you have it running 🙂 if you want the work coil to be the same, you can only add more capacity for lower frequency or less for higher frequency.

    The resonant frequency is the oscillating LC circuit, the chokes are not a part of this. The chokes will only change how stable your DC supply to the induction heater is. Too large chokes will however also present a significant voltage drop across them.

    Kind regards

  261. Hi again,

    You said that that for my AC to DC converter I need this ” KBPC2502 ” 25A , 200V bridge rectifier, but is this one ” KBPC2504 ” good too?

  262. Hi, I tried to make top electrical circuit and i made blow picture, now it doesn’t work, do you have see any problem with it?
    1) Transformer 12V (4A)
    2) Mosfet IRFP250N
    3) Resistor 470E/2W and 10K/1W
    4) Diode Bridge U860
    5) Zener doide 1N5349B
    6) Capacitor 1u/275V (MKP)
    7) Inductor are made from teriod(1mm copper and 48 twisted)

  263. Hi Mohsen

    There are several fatal flaws in your built that makes it not work.

    The first is the power supply, as you can read this circuit draws between 600 to 900 Watt, your power supply is only able to deliver 48 Watt. You need a higher voltage, preferably between 20 to 30 VDC and be able to deliver at-least 30 Ampere, else you will see the voltage being pulled down and risk that MOSFETs blow up from not switching properly.

    The veroboard will not be able to handle the currents in the DC supply that will be around 30 Ampere and in the resonant circuit you will have more like 100 Ampere circulating. Your board will just go up in flames, you need large solid copper connections.

    The components are correctly chosen and should work fine with the above corrections.

    Kind regards

  264. hi my friend
    thanks to answer me and i have some questions.
    do you think is enough i use power supply 24V dc with 15 ampere output or i use more?
    this picture is back of my board , do you think it’s better or i change it?

  265. Hi Mohsen

    You have to watch the voltage carefully, it will drop if the load is drawing too much current, if the voltage drops below 10VDC, you risk that the MOSFETs latch and explode.

    Tinned thin traces are not enough, maybe for very short runs, even my setup made with 2.5 mm^2 wires got warm in the short time it took to heat the M10 bolt.

    Kind regards

  266. Hi Mads,

    I’ve been messing around with a few different circuit designs thus far, and one trait shared between all of them is that past a certain voltage from the power supply, the current being soaked by the circuit reaches the maximum. Is there any reason that this occurs? (Just trying to make a semi-efficient circuit that still oscillates)


  267. Hi Tyler

    Do you have a current limited power supply? Do you mean that the voltage starts to sag at high currents?

    I am not too sure what you are asking about, could you describe it in more detail please.

    Kind regards

  268. Thanks for the response,

    It is a current limited power supply, for safety reasons haha. After increasing the voltage to a certain point, voltage stays constant around ~8.4 volts, and will sink as much current as it can. If I set the limit to 9 Amps, it will take 9 Amps.

    Also another issue is that when the circuit reaches a certain voltage/current, the diodes heat up, and I think they short the FETs, as after the diodes heat, the FETs no longer oscillate, and instead act as a short circuit.

    Apologies for the lack of clarity, finding just the right words is a little hard haha. My overall goal is to achieve maximum current across the tank circuit at high frequency, without damaging components. (I’m using a 2200 pF capacitor, and a 6.1 uH coil)

    Many thanks,

  269. Hi Tyler

    Your power supply is just too small, you need something that can supply atleast 1000 VA at 30 VDC, to avoid severe voltage sagging.

    You diodes are not the problem, voltages below 10 VDC are a hazard to your MOSFETs and you risk blowing them up from latching and shorting the DC bus. They can simply not maintain proper switching on/off at those low voltages.

    Kind regards

  270. Hi Mads,

    I’m going to look into a larger power supply to avoid voltage sag, but I am having troubles measuring the current through the tank circuit accurately. I have tried inserting a 1.5milli Ohm resistor in series with the tank circuit, but even that skews measurements too much. I do have an oscilloscope and multimeters at my disposal, but I just can’t get an accurate waveform/ measurement for the current through the coil.

    Warm Regards,

  271. Hi Tyler

    Be careful with the current sensing resistor/shunt, isolation is a problem, as you insert your measurement equipment into a high voltage loop in regard to ground. You risk shocking yourself or damaging your equipment.

    You can not use a normal multimeter to measure a 30 to 100 kHz current waveform, only your oscilloscope will work here.

    You should try to use a CT instead, you can easily make a 1:100 ratio one, 100 windings on a toroid ferrite core (your primary lead going through the center of it) and with a 10 Ohm burden resistor across it, you will have 10 V across the resistor when 100 A flows through the primary circuit.

    Kind regards

  272. But Mads,

    For high frequency work, if I wanted to know the current as well as the magnetic field produced by the coil. Would I be able to use a single coil to measure the induced voltage, and then integrate/ derive Phi to achieve the flux/ magnetic field? I’m having some difficulty with establishing a measure for magnetic field ~4 Mhz.


  273. If you want better efficiency, concentrate more on raising the Q-factor of the work coil tank circuit. Through careful component selection and construction methods I was able to achieve similar results shown here with only 130W of energy input (14.5VDC at 9A) with the load inserted into the working coil. At idle, with no load, my current draw was only 3.5A and yet my 9.5mm (3/8″ O.D.) working coil was too hot to touch in less than 60 seconds. Water coolant is absolutely necessary at all times. I haven’t actually measured the circulating currents yet but think about about how much current is required to do that.

    The working coil is 0.85uH. My MOSFETS and capacitors are where I spared no expense. MOSFETS are rated for 200V with RDS-on of <0.016 ohm. Capacitors are 1uF rated for 1200VDC/500VAC with only 0.005 ohm ESR. 10 in parallel yields a total of 10uF with ESR of only 0.0005 ohms. Oscillating frequency is 54 KHz. Project updates are being posted to my YouTube channel. And this is only a precursor for the ultimate intended purpose, creation of alloy metals needed for my NJM experiments (Non-Joulian Magnetostriction). That's where the fun really begins!


  274. Hi Tyler

    I do not have much experience in high frequency measurements, but maybe you can use a hall effect sensor to map the magnetic field? or simply just iron powder on a paper/plastic sheet to see where the field lines are.

    Kind regards

  275. Hi Zero

    I will watch your video later tonight, a high efficient IH without load might not be high efficient for induction heating, everything changes with different work pieces and even different frequencies can be needed to different kind of metals.

    I think you need some more space between your capacitors, for forced air cooling, the heavy busbar will do a good job as you can remove about 2/3 of the heat out through the leads/cooling the ends of a axial capacitor. But 1/3 will be radiated from the large circumference area, so your capacitor bank is prone to getting hot spots in the middle.

    Kind regards

  276. Hello again Mads

    Just wanted to share a picture of the transformer i scored for my power supply. I still haven’t assembled my tank circuit which might be a good thing because the mosfets I initially intended to use does not have high enough voltage rating for my upcoming power supply. I measured the rectified DC voltage today and it is 37V. The transformer is rated at 6kVA 🙂

  277. I am trying to understand each part and component of the circuit… I get the power resistors R3 and R4, I get the zener diodes D1 and D2, I get the mosfets Q1 and Q2 and I get the pull down resisistors R1 and R2. Now I’m trying to figure out the inductors L1 and L2. At first I thought of them as a DC choke but should they instead be viewed as a high frequency component made for the frequency of the tank circuit? In most cases somewhere around 100kHz and roughly 20A.

  278. Hi Mikael

    That is a proper transformer you have found there! When you have a working circuit, you should consider building paralleled bridges or switch to another topology. Roger here made a version with 4 bridges in parallel for a 4 kW version:

    The chokes are voltage storage/sources and are there to avoid heavy voltage sag during switching with a heavy work piece as a load. They are not a part of the resonant LC circuit.

    Kind regards

  279. Hello again

    Do you know if the admins at 4hv-forum are still active? I have tried to create an account there but not much response so far.

    I have managed to strip the welding power source of everything but the transformer and the large choke, so now I am trying to find suitable rectifier diodes that will be easy to mount and water cool. I tried to measure the choke and got 1.8mH with a Q value of 5 @ 1kHz, whatever that means. I will try and re-use the caps for the “primary side” of the choke and then buy new caps for the “secondary side” of the choke.

  280. Hi Mikael

    Write me a email through the contact form with the username and email you used for 4hv registration, then I will post it on the moderator forum for activation.

    Kind regards

  281. I have been toying with the “1000 watt 12-48 volt ZVS Induction Heater on and off for several months now. So far I think I have blown up about 4 of them. Mostly from my own mistakes. I think I have just about fool-proofed the system by now. So the next move is to upgrade my system from “Breadboard 1″ to Breadboard 2.
    A couple of points:
    -Why are so many of you guys trying to work at higher frequencies? I would think that lower is almost always better for heating metals, especially iron bearing metals (almost all forms of steel). My goal is to be able to heat a 1/2” square bar of mild steel to at least 2200 degrees F., so I can forge it. I will need to get the frequency down to about 30Khz or lower to do that.
    -This brings me to my second point:
    Mads, I think you mentioned something about metals and curie point or transition temperature somewhere in here. This phenomenon is unique to iron (and nickel-cobalt) bearing materials as far as I can tell.
    It has to do with the magnetic properties of iron. Until the iron becomes non-magetic, the little “dipoles”( magnets) in the iron attempt to switch back and forth to follow the magnetic field creating more friction (heat). But, once the material becomes non-magnetic, (at the transition temperature, which is around 1450°F,) they don’t do that anymore. So there’s less energy trapped in the material. If you have an ammeter (and EVERYONE should) you can see the current drop considerable when this point is reached
    -Third point:
    Many of you have asked about the size of the work coil for various reasons. But I don’t think most of you realize the large impact of changing to work coil diameter on a given piece of work material.
    Example: you attempt to place a large piece of steel into the work coil and the current goes up so high that you blow out the Fets and zeners. Well, if you increase the inside diameter of the work coil from, let’s say 1 1/2″ to 2”, you will get much POORER coupling and the current may well drop into a safe zone.
    I have shot several videos of my experiments, but non that I feel are ready for publication yet. But I did put up a webpage that summarizes my few successes and many failures and “learnings”. It is here:

    -I am still waiting for a bunch of 12 volt zeners so I can put some more boards back together. By the way, I’d never replace a Mosfet without also replacing the zeners.

    Why don’t those Chinese sellers ever put heat sink compound on their Mosfets?

    Pete Stanaitis

  282. Which components in the circuit are subjected to heat from losses so great they could use water cooling except the obvious mosfets and perhaps caps? Zener diodes or the other diodes? Power resistors?

    I think I have concluded that the chokes should preferably be made of powdered iron core as long as the frequency is kept well below 100kHz to avoid saturation. For my project I hope I can get down to about 40-50kHz.

    I would also like to share a little tip. I purchased the app recently and discovered there was a simulation of the royer circuit there made by a user. It’s great fun to toy around with but I suspect the readings you get aren’t very accurate but still educational.

  283. Hi Pete

    I agree that lower switching frequency is better for ferrous metals and also has a advantage when it comes to switching losses in the MOSFETs/IGBTs, tank capacitor and work coil.

    But if you want to heat non-ferrous metals you might have to change frequency to find a optimum working point for that particular type.

    Thank you very much for all your experimental notes that you included in your post 🙂

    Kind regards

  284. Hi Mikael

    There is only 3 components that should be water cooled, other components that heat up will just have to be dimensioned for the heat dissipation or have added forced air cooling.

    The 3 components are MOSFETs/IGBTs, resonant tank capacitor and the copper work coil.

    Be aware that many simulations never reflect reality, as the circuits does not contain all the parasitic inductance and capacitances that a real life circuit has. Even the slightest difference in wiring/layout resistance can dramatically change the currents flowing in a circuit and yet alone the predicted temperature rise. It can however be a good checked for logic faults and to see what kind of wave forms that could be expected in the circuit.

    Kind regards

  285. I have been trying to figure out the best way to construct the chokes but I am totally lost here. I don’t fully understand the function of it, and I want to create my own so I need to figure out the parameters I need. The design calls for a choke around 100uH. I have used the software on the following page to help me figure out a suitable toroid core and wire gauge.

    I like to over engineer stuff and I know that I would like something around 50kHz for my application and I hope to reach close to 2kW when I am finished. I know that my power supply will deliver close to 37VDC. That tells med that I would like a choke that can handle about 55A and the software tells me that a T400-60D core would work. I got a price quote for that core… 60 USD plus VAT and shipping! There is NO WAY the choke on the chinese ebay-boards can handle roughly 20-30A without saturation. So my question is: should the chokes be designed to saturate in order for this circuit to work properly?

  286. hello
    the only thing the inductors have to do is : a low resistance for dc but a high resistance for the working frequency , so the circuit will be able to oscillate .
    in other words , keep the frequency in the tank and not traveling to your powersupply

    if your inductor is too small ( inductance) : no oscillation
    if your inductor is too great , this means more resitance on the dc input and so limits the current

  287. Hello Mads.
    Regarding poor gate drive waveforms:
    I saw a video clip, I think from you, that talked about seeing a strange waveform—
    Here’s what I remember:
    You were going to give your circuit to a friend and as you were testing it, you found this strange waveform but you could not find out why it occurred. After quite a bit of trouble shooting, you discovered that the power leads were the cause. The leads were simply straight wires going from one place to another. You twisted them to cancel out stray inductance and the problem went away.
    Do I remember this correctly?
    I had a similar problem with an earlier setup.
    When I recently rebuilt my test setup, I did twist the power leads as much as I could and the gate waveforms improved considerably, but they are still not square waves.
    What can I do to further improve those gate drive signals?

    Pete Stanaitis

  288. Sorry to bother you guys again, but I think I have an answer to my own question about improving gate waveforms. It appears that, for the common Chinese 1000 watt ZVS induction heater, the higher the power supply voltage, the better the shape of the gate waveforms. By “better”, I mean that the gate waveforms become closer to looking like a square wave, which means to me that the Mosfets will spend less time in “linear” mode.

    I show some waveforms and discuss this on my latest revision of this page:

    Pete Stanaitis

  289. Thank you Mads and Andre. I still don’t fully grasp the function of the choke, but I have begun to create my own. I couldn’t find any T400-60D cores at a reasonable price so I bought a pair of T300-60D to start with. I figured if the ebay-models worked with those small chokes a 3 inch core probably would do worse. I couldn’t find any really thick copper wiring capable of handling 40-50A and also I think it would have been hard to wind a 6mm2 solid conductor around the toroid core so I made an (erroneous?) assumption that the wire may be optimized for the frequency of the tank circuit so I ordered a 0.5mm enamelled copper wire and made a wire that consists of 23 individually isolated 0.5mm solid copper wires creating an easy-to-bend wire close to 4,8mm2. The 0.5mm wire can conduct up to 68kHz with 100% skin depth.

  290. I meant to write that a 3 inch core probably WOULDN’T do worse than the smaller one. Estimated DC resistance of the wiring is somewhere around 10-20mOhm

  291. hi i have a question exactely What is the propouse of ferrite core round 130mH to supply the voltage 35vdc or choke what ever and if you have a teory for desing this or calculated and the 470 homs 5Watt power what is of 5 whatts or not 1/4whatt what is the propouse thanks

  292. Sir can you recomend type for fast diode, cause i used the FR207 the mosfet are burned
    And the capasitor bank is not mkp but it is 3 uf 400 volt


  293. Hi roa1212

    Read the latest 10 or so comments above your own for a discussion on inductor specifications.

    The 470 Ohm gate resistor has to be rated for 5 Watt power dissipation, this is because the MOSFETs are driven directly from the power supply to the resonant circuit and does not have its own dedicated power supply.

    Kind regards

  294. Hi alex

    You just used a normal power diode with a slow reverse recovery time, you need to use a diode that is called ultra-fast, this is a general description used on all electronic reseller sites. Find one that is not rated above 600 VDC, it is not needed and would make you chose a unnecessarily slow diode.

    What do you mean that the tank cap is not MKP? What kind of capacitors did you use? Not all types of capacitors will work in a high frequency high power resonant circuit!

    Kind regards

  295. Regarding the 470 Ohm resistor’s purpose:
    It is essentially in series with a 12 volt zener diode. When that zener diode conducts, any supply voltage in excesss of 12 volts will be dropped across it. In my case, I use a 48 volt power supply, so I could see 48 -12 = 36 volts across that resistor. If the zener diode was operating close to 1 watt, there would be about 80 milliamps in the circuit.
    P= E X I, so .36X .08 = approx. 2.8 watts. And that’s if there were NO spikes coming back toward the power supply.
    Just a note about the Mosfet gate: The gate draws almost no current. I can charge the gate by momentarily touching one hand to a 12 volt battery and touching the other hand to the gate and the device turns on completely. It stays on for quite a long time even when I remove my hand.
    I say this to point out that it’s the gate components, not the gate itself that will determine current flow in the 470 ohm resistor.
    I know I have not involved the 10K ohm resistor. It won’t have much influence on the discussion above.
    Does that make sense?

    Pete Stanaitis

  296. Ok, so the enameled 0,5mm wire I bought was a total nightmare to tin and solder so I am going to buy new wire, but before I go ahead and spend 70 USD on a few meter of copper wire I would like to know if there are any real advantages of 20+ conductors of 0,5mm wire instead of a single 3mm wire for this application?

  297. Hi Mikael

    A litz work coil will have a lot less losses than a single 3 mm wire, even at just 30 kHz the skin depth in copper is just 0.38 mm.

    But the conduction loss is not even the biggest issue, which is also why you wont see that litz can solve all the problems. You also have self induction issues and heat radiated from the work piece to the coil.

    Copper tubing is superior in almost any aspect, thin wall where almost all copper is being used to conduct current, sturdy, easy to bend and can be water cooled without problems.

    I explained many of these things in detail in my DRSSTC guide, on primary circuit design:

    Kind regards

  298. I’m afraid I didn’t explain that this wiring is for the choke. I wound the chokes with the wire I constructed but soldering was impossible so I decided to buy some other wire for the chokes, but since it’s a pulse load I was unsure whether to treat it like an AC or DC component, but I guess the current and magnetic flow never reverses. It just goes up and down, so I guess a single conductor will work just as good as many smaller ones.

    I put the original choke from my first ebay induction heater inside for comparison. Will post an update when I have received my new wire and rewound the chokes.

  299. Hi Mikael

    It should be treated as a DC component, but you are correct about the abrupt pulse current load that it will see.

    You need to remove the enamel from the wire in order to solder it, you can either scrape it off with a knife or remove it with a heavy acid, drain clog remover might be good enough to remove it, but be careful with chemicals, fumes etc, wear gloves and glasses.

    I would love to see you project with all details and pictures, please post it on the new forum, its a great help instead of single pictures in these comment tracks 🙂

    Kind regards

  300. to Mikael:
    I agree with Mads about removing the enamel. There is nothing wrong with your wire. The harder it is to strip the insulation, the better it will be for winding and for short circuit protection.
    I use a knife blade to carefully scrape the insulation and then follow up with about 80 grit sandpaper. If the wire doesn’t tin as easily as bare copper wire, then you simply haven’t gotten the insulation off completely. Sand some more and try again. One thought, though: is it possible that you have gotten aluminum enameled wire? (I have never seen any, but that doesn’t mean that it doesn’t exist). The difference in color should be obvious. I bought 118 feet of 14 Gauge enameled wire on Ebay for USD$25 recently. Why are you paying so much for a “few meters”?

  301. Hello
    I tried to make this heater but it works poorly at 12v and even then heated MOSFET. . When I grew up voltage to 30V both IRFP exploded.
    I used:
    Source 32V. secondary 24V (6mm x 3mm flat bar copper) bridge 400V/70A, 4x 10000mF / 63V filter capacitor.
    Working coil= copper pipe 10mm, 50mm inside diameter and 5 windings.
    MMC = 2 x 10mF / 300AC in series and 2 x 10mF / 300AC in parallel.
    2x ferrite 40mm x 24mm x 14mm, with 14 turns , 2.5mm.
    2x 250N X2 IRFP.
    10Kohm 2x / 2w .
    2x 470ohm / 5w .
    2x 1N5349B.
    2x RURP860.
    Where am I wrong? Why MOSFET burn? What is the reason?

  302. @ Pete Stanaitis :

    The only seller on ebay I could find that had enameled AWG 9 wire was Temco and the shipping was quite expensive so total cost for 37 feet was just above 70 USD

    I would have preferred AWG 8 or even thicker but then my T300-cores would have been too small unless I wound more than one layer. I think AWG 9 and a fully wound (single layer) T300-60D core will do for now. I actually looked at 4mm inner diameter copper tubing used in HVAC-systems and such for winding the choke but decided that would have been a little bit too insane, even by my standards 😉 (Also the equivalent cable area was too small)

  303. @ Mikael:

    I see that you already have a plan, but just in case:
    Why not go with 2 X 10 gauge wire?
    Here is 47 feet for USD$22.50 with free shipping:

    When I asked for heavy gauge magnet wire at a local motor rewind shop recently, they told me that they don’t use heavy wire any more for rewinding even the largest of motors. Instead, they use multiple strands of lighter gauge wire because it is easier to do.

    By the way, I also found 20 feet of 8 Ga. for USD$35.35 at Amazon;

    I use the search term “magnet wire” rather than “enameled wire”. We used to call it “Formvar” but I guess that’s an outdated term.

    Pete Stanaitis

  304. Thank you Pete!

    The Amazon seller actually is the same as the ebay seller I bought from. The shipping to Sweden will be the same whether I order from Amazon or ebay so no price difference. The ebay link is also the same seller. I paid almost 40 USD for the shipping.

    A pair of 10 AWG could work but it would mean that I would end up with more than one layer of wire on my toroid cores which I would prefer not to have.

    It is probably easier to wind a “litz wire” but to get the same conducting area as my 9 AWG wire I would have to put two layers on my core. That would probably work just fine but the longer length of wire would mean a larger resistance and therefore higher loss and more heat. Also, as you can see from my previous picture I ended up encapsulating the entire toroid core, trapping any heat generated inside the core. But it sure looks awesome! 🙂

    I think that a single wire and a single layer on the core is the best way to go if one would like to use it continuously as I plan to do.

  305. Hello,
    I built this circuit with:
    2x IRFP260N 200V 49A
    2x 470E 4W
    2x Z-Diode 1,3W
    2x BYV26E Si-Diode
    2x 10kE
    2x 2,2uF =4,4uF MKP 250V
    2x 64uH, but only 5A

    My circuit doesn’t work, I suppose it does not oscillate.
    For the Power Supply I use two 3S LiPo serial -> 25,2VDC 135A constant (157A Burst), 3,4kVA.
    I did not use a switch or something like that. First I conected the negative pole and then the circuit to the positive pole of the Battery -> spark to the plug of the battery connector -> molten XT60 plug and broken 30A Car-Fuse.

    What is the problem?
    Am I using wrong components?

    Kind regards

  306. Dear Mads

    I see your topic is so hot here, so I upload my question here to find some one who could give me some suggestion. if my issues are not suitable,please delete my issue. thank you.

    I run a TL494 chip to see the waveform output, however, the result is waving on the top and at bottom, look at the pictuers, why the result is this?
    Is there any way to modify the waveform?

    Thank you.

  307. Hi guys,
    can anyone tell me please how is the Power calculated? In the specification up the author mentioned that its 650 Watts. How can I calculate it for my own circuit and what is the way followed or equation used to know the exact power?
    I really appreciate your help and have a nice day.


  308. Peter Notebaart

    Hi Mads.
    I have two big GSM powersupplies each 27,2 volts.I have connected the powersupplies in serie .total 54,4 volts.This is to mutch for my inductionheather.I will make this voltage adjustable.Is this possible with several powertransitors parallel?Thank you.Have you an

  309. Hi Mads.
    I have build the Royer induction heater.It works fine.I took about 15 amps.When I use a variac to adjust the power then the fuse from the house is blown.How is it possible?
    Kind regards.Peter

  310. Hi Peter Notebaart

    The Royer oscillator is a really power hungry circuit and will take anything it can get. So when you lower the voltage, it will presumably just draw a higher current and that higher current is what would be tripping your mains fuses for the house.

    Kind regards

  311. Peter Notebaart

    Hi Mads,I see no more technical e-mails from other people on your site.Can you ask me why?
    Kind regards.

  312. Hi Peter

    I am not sure what you are asking me about? You do not see all the comments from other users or what do you mean by technical emails?

    Kind regards

  313. Peter Notebaart

    Hi Mads.
    About my last email.Technical emails are emails from other people from this site about induction heaters.I have another question.The induction heaters from Ali Express:Are they from good quality?
    Kind regards

  314. Peter Notebaart

    Hi Mads.
    I want to make 36 volt for my induction heater.I have 2 switching power supplies,eatch
    24 volt.2 in serie is 48 volt. that is to mutch.Can I make 36 volt with some power darlington transistors and a zenerdiode from 36 volt?
    Kind regards

  315. If you are talking about this induction heater, then 48 volts is not too much:,searchweb201602_2_10152_10151_10065_10130_10068_10344_10324_10547_10342_10325_10546_10343_10340_10548_10341_10545_10084_10083_10307_10059_10314_10534_100031_10604_10103_10142,searchweb201603_25,ppcSwitch_4_ppcChannel&algo_expid=30af7ebf-4d89-418d-9657-3ef2033c1b1a-21&algo_pvid=30af7ebf-4d89-418d-9657-3ef2033c1b1a&priceBeautifyAB=0
    Look at “Combo two”. It comes with a 48 volt 50 amp power supply.
    But you must be certain that your 24 volt power supplies can deliver at least 30 amps.

  316. Hi Peter

    As long as you respect the voltage rating of your MOSFET, with the circulating voltage in the circuit being 3.14 times higher than the input voltage and remember to account for switching spikes, so at-least 4 times higher is a okay safety margin.

    Kind regards

  317. hello mads. it is necessary to make that transformer ? can i just connect that to a 35vdc power supply? also, i recognized that this circuit is analog based circuit. so, what im going to do is to modify the diodes, zeners and resistors to simply connect it to an arduino for switching mosfets. how do you think of that ?

    im sorry if that questions seems funny to you.

  318. Hi Sam

    Yes, you need the inductors to have a voltage source to feed the inverter.

    The Royer is a self oscillating circuit, so you need to program the active driver circuit in order to make sure there is no shoot-through. The outputs of a Arduino can not deliver near enough current to drive the MOSFETs, you need driver ICs or another small MOSFET output stage to drive them.

    Kind regards

  319. Peter Notebaart


    I have a question.Ali Express sell an induction heater 1800 watts.69 $ 69.Can I place directly 48V on this machine?
    Kind regards.

  320. Peter Notebaart


    I have a question.Ali Express sell an induction heater 1800 watts.69 $ 69.Can I place directly 48V on this machine?perhaps can I place 24V first?
    Kind regards.

  321. Hi Peter

    Start with 24V first, 48V is already quite high for a Royer IH.

    It all depends on the voltage rating of the MOSFET/IGBTs. The switches will see around 3.5 times the input voltage + switching spikes/transients, so need to be sure that you have some head room to not destroy your silicon.

    Kind regards

  322. hello mads. i have a few questions for you. hope you can answer.

    1. can i use toroid inductor of 150uH but the current is 1.2A . is this acceptable?
    2. what type of diode you use to build the bridge recitifier?
    3. what parameter for the transformer ?

  323. Hi Sam Sodden

    1. You can only use such a small inductor if you design the IH to draw very little current, else the high current draw will drive the core into saturation and it heats up very fast.

    2. Use “ultra fast” diodes, these are usually the words used for it in the datasheets

    3. If you mean the input transformer for the 40VAC, you do of cause need a high enough VA rating so that you can supply the current needed. You need something around at least 800-1000 VA to get similar performance as mine.

    Kind regards

  324. Hi Sam

    That transformer looks good for the job, actually a lower voltage rating is not bad, it will however mean that the maximum output power is less.

    The Royer ZVS is a power hungry hog that will eat any power you throw at it, until it explodes, so the higher voltage the higher current will circulate, so 30 VAC will give you 30 VAC * 1.41 = 42 VDC and that means the MOSFETs sees pi times that, so you will have 133 VDC across the MOSFETs, leaves a good head room if you are using 200 V devices.

    Kind regards

  325. hello mads. i found an anomaly inside my simulation. the rating current for both inductors is 13A according to your scheme. however, from the smoothing capacitor, i measured rangely 150A-200A currents flowing to both inductors. the current rating of both inductors need to be up to 200A in order for the simulation to work. how is that possible? will this affect my real inductor of sec current nom of 25A from the transformer?

    2nd : i found a wire of CSA 1.5mm^2 . so, to wind the inductors, should i tear the pvc coating of the wire or keep it?

    thank you.

  326. Hi Sam

    I removed the old schematic you posted.

    Your simulation sounds like it is too ideal, you are properly missing circuit resistance and stray inductance/capacitance, and thus you are getting too high current flowing in the simulation, as its trying to drive very low impedance loads.

    Kind regards

  327. hello mads. thanks for the reply. regarding to those matter, i think maybe it is because of the inrush currents of the transformer. as i measured longer, the current flowing before the inductors dropped from 200A to 10A. do you install any circuit breaker to your experiment? how to choose a circuit breaker? im afraid that by not doing so, i will do massive damage to my components. i have a few components to be listed :

    1. hammond toroid transformer 60v 750VA
    2. irfp250mpbf
    3. 1uF Illinois cap 275v mkp x2
    4. 1n5349bg zeners 12v 5w
    5. few 3w resistors
    6. mur880eg fast 8A, 800V
    7. gbpc2502 bridge rect 200v 25A

    what do you think of my component listing?

  328. Hi Sam

    It all looks good, all your components are within rating, but the output voltage of your transformer is very high, you will most likely exceed the voltage rating of your IRFP250.

    60VAC rectified is 85VDC and the MOSFET will see pi times that, so 265VDC across a 200V rated MOSFET.

    Circuit breakers will not be fast enough to react on a over-current/short circuit, your MOSFETs will always blow up first. The only protection you can use for silicon is with large IGBT bricks that has a 1 ms rated high short-circuit current and you can use silicon specific fuses in front of that.

    Circuit breakers have never been meant to protect your circuit, they protect your installation up to the device, to protect the cable from melting/burning.

    You need local fuses with selectivity in regard to the main fuse, so that your local glass fuses goes before the circuit breaker.

    Kind regards

  329. hello mads. im sorry. it is a parallel transformer with 30VAC 25A each. also now im ordering 6x 6500uF 100V electrolytic capacitor for the smooting circuit. hope everything will going well.

    what wire do you use for your inductors ? it is just a common CSA 1.5mm^2 electrical wire? should i tear or keep the PVC?

  330. Hi Sam

    Okay, no worries about the transformer then. I just used regular wire that has the rating to carry the DC current expected, do not tear off the insulation.

    Kind regards

  331. hello mads.

    based on your video, it’s unclear of how you powered on and switch off this circuit. mind explaining that?

  332. Olá,

    Necessito fazer um trabalho da Universidade onde construo um aquecedor indutivo ZVS com os dados abaixo, poderia me auxiliar?
    V entrada = 220Vca
    F entrada = 60Hz
    Fsaida > 50Khz
    Potência = 300 a 1000W

  333. Hi Sam

    I simply just turned on/of the 230 VAC supply to the transformer. I am aware that the large electrolytic capacitor bank will make the DC voltage fall slowly, but I have not had any defects from switching on/off, so even if there was a cross-conduction error at switching off, the energy available to short-circuit is too small do cause damage.

    Kind regards

  334. Hi Marcos Lopes

    It sounds like you have this as an assignment. I suggest you learn how to do this yourself as that is the whole idea with taking a engineering degree.

    You can not afford for me to design one for you. Use the material that is available and do your own research.

    Kind regards

  335. Hello mads, what are the effect of using this same circuit, but using bigger work heating coil?

    Also, I’ve noticed that you use a single core wire to wind your capacitor bank to your heating coil. What type of wire is that?

  336. Hi Sam

    A larger coil will have a lower resonant frequency, but it will not necessarily have the power to heat the object that you want. Coupling gets smaller with a bigger coil.

    I used regular stiff copper wire for house installation, stripped from its insulation. It is a solid 2.5 mm^2 wire.

    Kind regards

  337. Mads, is it wise for me to do the same but with multiple core 4mm^2 wire? I’m worried about the max current each core can handle.

  338. Hi Sam

    Remember there is a difference between the “low” DC supply current from your capacitors through the chokes and MOSFETs and the “high” resonant current in your LC circuit.

    At higher frequencies there is something called the skin effect, that current simply can not penetrate the copper very deep, so using thick conductors become useless at high enough frequencies. Check out my wire size table here on the site.

    Kind regards

  339. Thank you mads for that opinion. I’ve completed this circuit using my components. But once i switch on, my house’s circuit breaker tripped. At my transformer, i used time delay fuse of 10A. I double checked my connection thrice, but can’t find any fault.

  340. Hi Sam

    The inrush current to the DC bus capacitors looks like a short-circuit when they are completely discharged. You could try to add a NTC inrush limiter in front of the transformer.

    Kind regards

  341. hello mads, i’ve finished constructing this induction heater. i have encountered a problem.

    there is voltage after the smoothing capacitor which is 42vdc. but there is no voltage after the two mosfets. and my 470ohm resistor heats up really high that it feels like burning when touch. why is this happening? is it because of my resistor?

  342. Hi Sam

    You can only measure the output of the inverter with a oscilloscope, a regular multimeter is not fast enough to read out the high frequency alternating current. If you have a signal output that goes to +10 and -10, a RMS multimeter will just show that as 0.

    Check your circuit for errors, double double check everything for wiring errors. The 470R resistors are gate resistors and should not get so hot its smoking.

    Kind regards

  343. wow, you just kept on answering questions for over 5 years. are you a teacher in real life?

  344. Hi 3RL

    I work as a lead engineer in programming, so once in a while I do have new engineers straight from school that I teach the ways of doing things in the company, but not a teacher in the traditional way 🙂

    I got so much help from people all over the world when I first started with high voltage, so I always tried to give back to others as well.

    Kind regards

  345. Hi Satman

    You are right, that is an error in the schematic. I wrote the correct 3 uF value in the specifications table and further down in the text, exampled here: “The work coil is made from 5 turns of 8 mm copper tubing, giving approximately 0,477 uH. The MMC consists of 9 parallel strings of 3 in series Rifa 1 uF / 275 VAC MKP X2 capacitors for 3 uF. This gives a resonant frequency calculated to about 133 kHz.”

    Thank, I will update the schematic.

    Kind regards

  346. Hi there,
    I’m trying to build an induction heater for very small, hollow objects (diameter around 1mm), wall thickness around 150µm. So far, I’ve already built some heaters with center tapped coil, that work very well for bigger objects, but they fail to heat to the small parts.
    Could you please give me some advise on how to modify the circuit to get it to work properly? (like diameter and turns of working coil, tank capacity and so on)

    Thanks in advance,


  347. what about using 4 mosfets in parallel (irfp260) for each oil output each with there own set of resistors and Zener?
    i seem to understand that should help share load between them somewhat
    i have 9 – .33uF MKP for tank – measures just over 3uF. should i get more for high uF?

    on the choke coil, does the diameter matter as far as the mag field needed. i have some 20A 100uF coils that are about 1″ outside diameter.

  348. HELLO;
    1- I have 2 24v 13a power supplies. I can get 48v 13a or 24v 26a power by connecting in series or parallel. Do you mind if I do this? Which is correct, series or parallel?

    2- Can you suggest another component that I can use instead of rurp880?

    3- I will do water cooling in the whole system. Is it correct to use pure to prevent corrosion?

    4- Is it necessary to make current protection for Mosfed?

    5- I have 10 capacitors of 0.33 uf 1200v. Would you mind if I make 4 uf 1200v 1 piece or 1 uf 1000v 4 piece instead, which one would you recommend?

    2x ırfp260n
    10x 0.33 uf 1200v mkph capacitors
    1n5349b 5w zener
    470r 2w
    10k 1w
    rurp880 ?????

  349. Hi Emre

    1: The higher voltage, the higher current a ZVS driver like this, will draw. So for that current rating, 24V is properly maximum without power supply shutting down from over-current.

    2: Any fast / ultra-fast diode with same ratings as rurp880 can be used.

    3: You can use pure water if its just for short runs, if you need long time stable unit, use coolant corrosion inhibitor. Water will however get polluted over time from ion’s wandering.

    4: These oscillators are prone to failure, due to their simple self-oscillating nature. Implementing MOSFET over-current protection is properly making this circuit more complex than just building a regular half- / full-bridge with proper over-current protection.

    5: Both 1 uF and 4 uF would work, try some different setups and see what works the best 🙂

    Kind regards

  350. Hello again

    I’ve been working on this project for a long time, experimenting when I have free time. My only issue with this system is the hot parts. The 470r 2w resistors heat up immediately and explode. Instead I connected a 470r 5w stone resistor. I updated the 10k resistors to 10k 2w. without water cooling the capacitors are overheating.
    As a result, I cooled the coils and capacitors with water. I made a special aluminum heatsink for the mosfeds and connected it to the water cooling line. My next goal is to turn the whole system into a closed box and make it air-cooled.
    capacitors still get hot despite this much cooling process.
    Could the components I use be insufficient?
    System information is as follows.

    my power supply is 24v 26a
    working frequency 128khz
    ırfp260n x2
    0.33uf 1200v x10 caps
    200uh ınductor
    1n5347b 5w zener diode
    by228 fast diode
    10k 2w resistor
    470r 5w resistor
    work coin 3turn 6mm

  351. Hi Emre

    Everything looks fine, except your work coil. You only have 2½ turns where I use 4½ turns. So your resonant frequency and peak current in the primary circuit is much higher. You have higher losses at higher frequency and higher current. That is why you have all this heat dissipation.

    Kind regards

Leave a Reply

Your email address will not be published. Required fields are marked *

The maximum upload file size: 10 MB. You can upload: image, document, spreadsheet, text, archive, other. Drop file here

Scroll to Top